View Full Version : Question for GFI advocates:
I was just running some number using GFI, and I came across this dilemma.
I was looking at NC #7 vs Casino #3
Here are the #'s
I'm assuming I make $3,000,000/hr
NC #7
IpH increase - $352,917
Cost - $1,449,023,034
Upgrade Time - 108:36 (I have Loaded Dice)
GFI = 597
ROI = 4,105 hrs
Casino #3
IpH increase - $ 61,875
Cost - 167,334,000
Upgrade Time - 51:50
GFI = 575
ROI = 2,704 hrs
Now, do you GFI advocates believe that NC #7 is the better upgrade?
The investment in the Casino gives you approximately a 50% better return on your investment. And since you are forced to save for either investment based on your income, the argument that ROI favors "cheap" investments is not valid here.
I would like to hear the arguments for the NC #7 upgrade and simply because it has a larger IpH increase is not a valid argument because it does not take into account cost which in the investing world is half the equation.
Dipstik
10-16-2012, 08:54 PM
I upgrade the prettiest buildings!
Euchred
10-16-2012, 09:00 PM
I upgrade the prettiest buildings!
Me too I love the pink highlights on the palm hotel.
But seriously the palm hotel is aesthetically pleasing.
GFI does seem simple and elegant as it seems to include all the necessary components for calculating an optimal economic plan. But one of my main contentions, and has been for quite some time, is that too much stock has been put into upgrade time. I think it is a very simplistic assumption to believe the upgrade time, IpH and Cost all have linear relationships.
Here is an example using a $500,000 income.
Modern Loft #3
IpH increase - $4,125
Cost - 8,366,700
Upgrade Time - 20:15
GFI - 112
ROI - 2,028
Beach Nighclub #4
IpH increase - $3,850
Cost - $4,657,463
Upgrade Time - 39:32
GFI - 79
ROI - 1,209
Now, the Modern Loft costs 80% more for an 7% increase in output, but it does take twice as long to build. To me, GFI penalizes the Beach Nightclub far too much, $$/hr/hr would penalize it even more, since it doesn't take in cost at all.
The reason that should make sense is at your income (you're at $3m? Good work.) the 2.1% IPH increase is fairly marginal. It currently only takes you about a minute and 20 seconds to make the IPH gain (and a little over 2 days to save for the upgrade).
The NC, on the other hand, is a sizeable 11.8% increase to your IPH.
At that IPH the casino upgrade could be used in the middle of the NC save because it is so cheap. So is it a better upgrade? Depends on your definition of good. My definition of good is adding to my IPH in a significant manner, so I'd call the NC the main upgrade to be saved for and the casino something to pass the time with. If you define good as something that is paid off quickly then the casino is the better upgrade - just realize that the building increased your daily income from $72M to $73M whereas the NC increases your daily income from $72M to $80M.
I still believe GFI needs a modifier on save time, but the above should show you why the Casino, in that instance, isn't a game changing upgrade.
Sorry I'm new to this but what does GFI mean? Thanks
The reason that should make sense is at your income (you're at $3m? Good work.) the 2.1% IPH increase is fairly marginal. It currently only takes you about a minute and 20 seconds to make the IPH gain (and a little over 2 days to save for the upgrade).
The NC, on the other hand, is a sizeable 11.8% increase to your IPH.
At that IPH the casino upgrade could be used in the middle of the NC save because it is so cheap. So is it a better upgrade? Depends on your definition of good. My definition of good is adding to my IPH in a significant manner, so I'd call the NC the main upgrade to be saved for and the casino something to pass the time with. If you define good as something that is paid off quickly then the casino is the better upgrade - just realize that the building increased your daily income from $72M to $73M whereas the NC increases your daily income from $72M to $80M.
I still believe GFI needs a modifier on save time, but the above should show you why the Casino, in that instance, isn't a game changing upgrade.
See in this case the Casino is not something to pass the time with, because it's savings neutral. You cannot use the cheap upgrade argument in this case. It takes a little over 2 days to upgrade and 2 days to save for the cost. So at the end of the upgrade, I will have the same amount of money saved.
I keep using the argument that the flip side of ROI is yield. If an investment takes 5 years to pay off, the yield on that investment is (appox) 1/5 or 20%. And when I'm earning x dollars per day and I'm approached with 2 investments that are not mutually exclusive, one costing 2x and the other costing 10x, and the 2x investment yields double the 10x investment, I'm not sure how the 10x investment is better.
SiennaSharpe
10-17-2012, 09:10 AM
Sorry I'm new to this but what does GFI mean? Thanks
See here: http://www.funzio.com/forum/showthread.php?40719-Crime-City-Forums-Beginners-Glossary
See in this case the Casino is not something to pass the time with, because it's savings neutral. You cannot use the cheap upgrade argument in this case. It takes a little over 2 days to upgrade and 2 days to save for the cost. So at the end of the upgrade, I will have the same amount of money saved.Whoa, government math alert. Oddly, I think your explanation actually argues in favor of the NC being a better investment, but I think we need to fundamentally address something you're not grasping.
There's a cost of capital associated with any amount of liquid cash or any investment. Saying that it takes 2 days to save but 2 days to upgrade so it is revenue neutral is, at best, misinterpreting what's happening. Any amount of spent money is a sunk cost. For something to be neutral you'd need to offset it or have the investment pay off within the time frame you're looking at. The break-even point for the Casino is 112 days, far longer than the 20 save days. So, you are delaying the NC7 upgrade - the Casino is not a neutral investment within the upgrade period.
To illustrate this, I plugged these two investments into a spreadsheet and calculated 200 days of these two investments.
I assumed we already had enough cash on hand for the Casino upgrade. I also assumed we kept collecting the $3M/hr. I also assumed we didn't lose the IPH during the upgrade (you'll see why in a minute this doesn't really matter).
So, on day 1, Scenario 1 started with a C3 upgrade and Scenario 2 started with a save for the NC 7.
On day 3 the Casino started paying out.
On day 20 the NC started it's upgrade.
On day 25 the NC finished it's upgrade.
On day 30 C3 had an ROI of -.76 (note, this is actual ROI, not "CC ROI") and NC7 an ROI of -.96, so far the Casino is a more efficient upgrade. C3 had an IRR of -7.5% and NC7 an IRR of -21.7%, so far the Casino has returned, or yielded, more money
Day 37 is the start of what's actually interesting. NC7's IRR becomes higher than C3's (-4.7% to -5.3%, respectively)
Between day 37 & 38 the Net Present Value of the investments is higher for the NC7 than C3 ($4.02B vs. $3.89B).
On day 42 the NC's IRR hits 0%. The Casino's IRR is -4.19%. ROI for NC: -.89, Casino: -.65.
On day 115 the Casino ROI and IRR hits 0 and 0%. NC ROI: -.47; IRR: 6.86%. So when the Casino breaks even the NC is already yielding 6.86%. Even if we had taken into account lost income due to upgrade times you're only looking at a .01% of less yield (which is why it doesn't really matter).
Just for fun, on day 200.
C3
NC7
Cash
$21.3B
$48.6B
Gain
$0.293B
$1.49B
ROI
0.75
0.03
IRR
0.62%
6.92%
As you can see, ROI is poor indicator of which is the better investment.
http://imageshack.us/a/img31/4936/c3nc7cash.jpg
http://imageshack.us/a/img268/1028/c3nc7roiirr.jpg
Ramshutu
10-17-2012, 09:54 AM
The numbers tell you that if you had a straight one-off choice between Casino and NC upgrades, over the upgrade and saving cycle, the NC is better.
What it doesn't say, and why you have misgivings, is whether the casino+something else+something else upgrade and save cycle is greater than the NC upgrade and cycle, in the situation where they take the same amount of time.
Essentially, GFI doesn't accout for the economic activity that could occur in the time difference between upgrades. That cannot easily be expressed in an equation (or at least not by one simper than the Reinman tensor).
I am doing some work with my simulation to work out whether I can program an algorithm that factors this in, isn't complex enough that it can beat Kasparov in a game of chess, and not slow enough that the universe will end before an answer.
I am doing some work with my simulation to work out whether I can program an algorithm that factors this in, isn't complex enough that it can beat Kasparov in a game of chess, and not slow enough that the universe will end before an answer.What about taking the next x (5? 10?) upgrades below it, calculating the expected returns as (IPD gain)/cost and use either the average or some function that combines them (Sigma (gains)/Sigma (costs), not sure that makes more sense) as a cost of capital. Plug that number in as the discount rate of a Net Present Value over some period of time (maybe the break even point?) and use NPV to sort.
Ramshutu
10-17-2012, 10:10 AM
What about taking the next x (5? 10?) upgrades below it, calculating the expected returns as (IPD gain)/cost and use either the average or some function that combines them as a cost of capital. Plug that number in as the discount rate of a Net Present Value over some period of time (maybe the break even point?) and use NPV to sort.
The simulation orders things using a list sorted by xxx (where x orders by algorithm).
I'm planning to allow it to cross reference stacked upgrades, provided I can find a way to simplify the logic of it. N!=NP solutions are no fun to implement.
Realistically though, there is no problem in the world that can't be solved by adding another level of indirection.
I'm having an incredibly hard time making you guys see any of my points....
Saying that it takes 2 days to save but 2 days to upgrade so it is revenue neutral
I didn't say revenue neutral, I said savings neutral. What I meant was this isn't a "cheap' upgrade to do while saving for a more expensive upgrade. I, in fact, would include this as one of the expensive upgrade (albeit at the low end of expensive) to actually save for because you don't end up with net savings at the end of the upgrade as you would if you choose to upgrade a Pizza Parlor say.
The rest of your argument assumes that these are mutually exclusive investments, which in fact they are not. In your argument, the cost of the investment doesn't factor into the decision process, it merely delays where you see the NC surpass the Casino in earnings. Whether the NC cost $1b, $3b, or $10b to upgrade, you would eventually conclude on longer and longer time horizons that long-term the NC is the better investment. That argument seems extremely flawed in a NON-mutually exclusive investment world.
This will be my last attempt at an explanation. If I'm not able to convey my point with this illustration, I will have to respectfully bow out of the ongoing discussions on economy.
I am going to use the NC and Casino upgrades and a $3mm/hr income as an illustration. Try to think about the point I am trying to make and please do not argue semantics.
NC
Cost - $1,449,023,034
Addt'l IpH - $352,917
Time to save for upgrade - 483 hrs
Casino
Cost - $167,334,000
Addt'l IpH - $61,875
Time to save for upgrade - 55.78 hrs
Upgrade time - 52 hrs
Now, instead of investing in the NC, let's assume I can find 8.65 ($1,449m / $167m) investments exactly like the Casino. Then every 55 hrs (my time to save), I choose the Casino/Casino-like upgrade. The result would be an increase in IpH of $535,218 ($61,875 * 8.65) vs the NC of $352,917.
I hope I have made my point clear enough so that you can at least argue against what I am actually trying to prove and not misconstruing my arguments.
The numbers tell you that if you had a straight one-off choice between Casino and NC upgrades, over the upgrade and saving cycle, the NC is better.
What it doesn't say, and why you have misgivings, is whether the casino+something else+something else upgrade and save cycle is greater than the NC upgrade and cycle, in the situation where they take the same amount of time.
Essentially, GFI doesn't accout for the economic activity that could occur in the time difference between upgrades. That cannot easily be expressed in an equation (or at least not by one simper than the Reinman tensor).
I am doing some work with my simulation to work out whether I can program an algorithm that factors this in, isn't complex enough that it can beat Kasparov in a game of chess, and not slow enough that the universe will end before an answer.
Thanks Ram, yes that is part of my argument....and is what I tried to illustrate with the above example
Ramshutu
10-17-2012, 10:26 AM
Thanks Ram, yes that is part of my argument....and is what I tried to illustrate with the above example
I'm trying to work out how to do this in the simulation without having to compare every single possible combination of all other buildings :)
I'm trying to work out how to do this in the simulation without having to compare every single possible combination of all other buildings :)
It would be nice if you could implement that, but even without implementing it, I still believe that ROI is the most important factor in considering an investment. Yes, it needs to be used properly (and I believe GFI attempts to do this), so that you aren't stuck focusing on cheaper upgrades, but as I keep harping on, the inverse of ROI is earnings yield (or return). And the main 2 inputs into evaluating an investment are risk and return. In CC, risk (variance of return) is zero, so all we are left with is return. Again, this has to be implemented properly, but I think to ignore it entirely is at your own cost.
It would be nice if you could implement that, but even without implementing it, I still believe that ROI is the most important factor in considering an investment. Yes, it needs to be used properly (and I believe GFI attempts to do this), so that you aren't stuck focusing on cheaper upgrades, but as I keep harping on, the inverse of ROI is earnings yield (or return). And the main 2 inputs into evaluating an investment are risk and return. In CC, risk (variance of return) is zero, so all we are left with is return. Again, this has to be implemented properly, but I think to ignore it entirely is at your own cost.Yield is another term for IRR, which doesn't track with ROI (since ROI is a measure of the efficiency of the investment). They're two separate things.
And, like I showed above - taken in a vacuum - yield for a NC7 blows away C3 after 37 days.
Also, return is lower for the C3...so I'm not sure what point you're trying to make with risk and return. You're making the point for GFI to be used over ROI again.
Yield is another term for IRR, which doesn't track with ROI (since ROI is a measure of the efficiency of the investment). They're two separate things.
And, like I showed above - taken in a vacuum - yield for a NC7 blows away C3 after 37 days.
Also, return is lower for the C3...so I'm not sure what point you're trying to make with risk and return. You're making the point for GFI to be used over ROI again.
I don't agree that IRR equals yield....
Yield = income return on an investment (roughly = income / investment)
IRR = Discount rate used for an investment to give an NPV of zero. The IRR will change depending on the time horizon chosen.
The yield of an investment in CC is static...you invest x, it returns y, y / x is your yield...yield is NOT time dependent, so saying that after 37 days the yield of the NC blows away the Casino isn't correct.
I'm trying to state one of the most basic ideas in investing. If I invest $100, and it returns $20 per year, my yield/reutrn is 20%/year. Simple, done. If another investment costs $200 and returns $30 per year, it returns 15%/year. The return on Asset A is greater then Asset B. Simple, end of story.
I don't agree that IRR equals yield....
Yield = income return on an investment (roughly = income / investment)
IRR = Discount rate used for an investment to give an NPV of zero. The IRR will change depending on the time horizon chosen.
The yield of an investment in CC is static...you invest x, it returns y, y / x is your yield...yield is NOT time dependent, so saying that after 37 days the yield of the NC blows away the Casino isn't correct.
I'm trying to state one of the most basic ideas in investing. If I invest $100, and it returns $20 per year, my yield/reutrn is 20%/year. Simple, done. If another investment costs $200 and returns $30 per year, it returns 15%/year. The return on Asset A is greater then Asset B. Simple, end of story.Look, that's a measure of efficiency, not how much money you're making.
If you have a $100 investment with a 10% yield that isn't better at the end of the year than a $1000 investment with a 5% yield. It's a factor, but it's silly to blindly assume that's a good metric. You need to put it in context with another factor. That's one reason why NPV is generally used alongside it, they both look at 2 sides of one coin. One prioritizes a small, efficient investment whereas one prioritizes a large investment (regardless of yield/ROI). Taking only one of them into account is a poor choice in the business world as well in CC for that very reason. If your objective is long term growth, it's not a very good metric. As shown in your example, after 10 years the investments are equal (assuming 0 cost of capital). After that, investment B is a better investment "end of story". Arbitrarily cutting it off in the short term is, excuse the pun, short-sighted - hence why you can't just blindly follow ROI.
Look, that's a measure of efficiency, not how much money you're making.
If you have a $100 investment with a 10% yield that isn't better at the end of the year than a $1000 investment with a 5% yield. It's a factor, but it's silly to blindly assume that's a good metric. You need to put it in context with another factor. That's one reason why NPV is generally used alongside it, they both look at 2 sides of one coin. One prioritizes a small, efficient investment whereas one prioritizes a large investment (regardless of yield/ROI). Taking only one of them into account is a poor choice in the business world as well in CC for that very reason. If your objective is long term growth, it's not a very good metric. As shown in your example, after 10 years the investments are equal (assuming 0 cost of capital). After that, investment B is a better investment "end of story". Arbitrarily cutting it off in the short term is, excuse the pun, short-sighted - hence why you can't just blindly follow ROI.
I have never said to blindly follow ROI, but all of your arguments look at 2 investments in a vacuum.
If I have 2 investment choices that are NOT mutually exclusive (therefore not in a vacuum), and both of the those investment choices are more then my current cash on hand, I will almost always choose, to save for and then invest in, the higher yielding (income/cost) project first. If we don't agree on this point, then we're just going to have to agree to disagree.
Ramshutu
10-17-2012, 12:30 PM
There are some big issues when comparing real world economy, and the CC economy:
- Cash on hand does not earn interest. If money is not spent, it is a waste of resources; and needs to be offset with some real benefit somewhere else.
- There are no running costs, no overheads, no staff pay. No contingency issues.
- There is no risk. This is like an infinite term bond.
- You can only make one investment at a time. (Two if you include building)
The ideal economy path is to have the largest $/hr/hr gain over time. Pure ROI, even subtly biased ROI doesn't work, at least by itself, as by game design, buildings with high ROI have have low $/hr/hr gains, and vice versa. More importantly, because of the cash-on-hand-being-worthless issue the whole purpose of ROI, to maximise available cash by maximising upgrade cost efficacy is unhelpful and even disingenous.
There is not really a right answer, or at least no one has one yet. Because while you know what upgrade has the highest $/hr/hr, and which of those upgrades you can actually afford, there is a balancing act between choosing a lower $/hr/hr upgrade that you can afford, and saving for a higher $/hr/hr upgrade that you can't. This is what GFI is a measure of, it's a measure of where "that balance" occurrs.
In that respect, in my opinion (and I don't mean to sound arrogant), GFI is the best "like for like" comparrison metric, as it allows you to chose the specific building that has the greatest $/hr/hr increase to your economy which essentially a measure of your economic growth. It can even be represented as an hourly, or daily %increase to your economy, like measuring GDP.
Now, this is not to say it is optimal for the economy as a whole, because that direct comparrison does not deal with stacking of different upgrades; but as no other metric or measurement will give an accurate numeric comparrison of the growth of your economy it's better than the others.
What is yet to be seen, is if there is a more complex algorithm that will allow comparrisons of not just one building to another, but one building to all other combinations. I think that there is given the example murf stated; but I just haven't thought of it.
I think that's my whole problem with murf's case - it relies on an imaginary strawman argument that two upgrades can be compared using ROI but can't be compared using any other metric. For us to simplify two defined investments enough to comprehend we have to look at them in a vacuum or with very limited alternatives (Ram, you might be different, but I'm just a simple man). With a simulation we could calculate more and actually have a blueprint for the best way to upgrade.
Thus, I can't wrap my head around his asking the question of which is a better investment but only relying on one datapoint and rejecting the others (because the two investments are now incomparable). Without that simulation I don't think I can quite get to theoretical/fantasy realm murf's argument lives in.
Ramshutu
10-17-2012, 12:56 PM
I'm a simply person, I like simple answers. But when I am interested in something, I HATE not knowing the exact answer! At the moment, I can explain how I would chose between the NC and Casino without any issue, but I can't do it with a quantative argument that would work in an algorithm
Murf:
I can put your concept into the simulation and see what answers it gives in comparrison to the other metrics, but I need you to describe it precisely in terms of a set of steps.
For example, the GFI Algorithm is:
1.) Calculate GFI for all available upgrades.
2.) Upgrade the building with the highest GFI.
Another example, is the "$/hr/hr" Algorithm which is:
1.) Calculate $/hr/hr of the upgrade.
2.) Find the building with highest $/hr/hr
3.) If this building is not affordable, discard it, and repeat from 2.
If there is a calculation you need to perform, can you please let me know the exact maths and I will le-plug it.
I will also let you name the algorithm, like that guy Nagle (he is in fact an a***hole given how much I have been strugling with it the last couple of days).
There are some big issues when comparing real world economy, and the CC economy:
- Cash on hand does not earn interest. If money is not spent, it is a waste of resources; and needs to be offset with some real benefit somewhere else.
- There are no running costs, no overheads, no staff pay. No contingency issues.
- There is no risk. This is like an infinite term bond.
- You can only make one investment at a time. (Two if you include building)
The ideal economy path is to have the largest $/hr/hr gain over time. Pure ROI, even subtly biased ROI doesn't work, at least by itself, as by game design, buildings with high ROI have have low $/hr/hr gains, and vice versa. More importantly, because of the cash-on-hand-being-worthless issue the whole purpose of ROI, to maximise available cash by maximising upgrade cost efficacy is unhelpful and even disingenous.
There is not really a right answer, or at least no one has one yet. Because while you know what upgrade has the highest $/hr/hr, and which of those upgrades you can actually afford, there is a balancing act between choosing a lower $/hr/hr upgrade that you can afford, and saving for a higher $/hr/hr upgrade that you can't. This is what GFI is a measure of, it's a measure of where "that balance" occurrs.
In that respect, in my opinion (and I don't mean to sound arrogant), GFI is the best "like for like" comparrison metric, as it allows you to chose the specific building that has the greatest $/hr/hr increase to your economy which essentially a measure of your economic growth. It can even be represented as an hourly, or daily %increase to your economy, like measuring GDP.
Now, this is not to say it is optimal for the economy as a whole, because that direct comparrison does not deal with stacking of different upgrades; but as no other metric or measurement will give an accurate numeric comparrison of the growth of your economy it's better than the others.
What is yet to be seen, is if there is a more complex algorithm that will allow comparrisons of not just one building to another, but one building to all other combinations. I think that there is given the example murf stated; but I just haven't thought of it.
I agree with most of what you are saying Ram. My main problem with GFI (I do think it's in has great value) it that it gives too much weight to upgrade time and not enough to cost of the investment. $/hr/hr gives no consideration to the cost of the investment, which to me makes it useless in decision making.
My background is trading Fixed Income Derivatives including some very complex option securities and what I have found in my career, is that rather then try to create an elegant solution which requires a variety of assumptions embedded in the volatility models, I have always opted for the simpler solutions where the limitations of the model are easily understood and able to be worked around.
This is the main reason for my strategy of ranking my upgrades by IpH, and then choosing the best ROI locally. This does not mean I choose Pizza Parlors over NC upgrades, but it does mean that I choose that Casino #3 over NC #7. This is a simple solution, where I know the limitations to my strategy immediately.
I'm a simply person, I like simple answers. But when I am interested in something, I HATE not knowing the exact answer! At the moment, I can explain how I would chose between the NC and Casino without any issue, but I can't do it with a quantative argument that would work in an algorithm
Murf:
I can put your concept into the simulation and see what answers it gives in comparrison to the other metrics, but I need you to describe it precisely in terms of a set of steps.
For example, the GFI Algorithm is:
1.) Calculate GFI for all available upgrades.
2.) Upgrade the building with the highest GFI.
Another example, is the "$/hr/hr" Algorithm which is:
1.) Calculate $/hr/hr of the upgrade.
2.) Find the building with highest $/hr/hr
3.) If this building is not affordable, discard it, and repeat from 2.
If there is a calculation you need to perform, can you please let me know the exact maths and I will le-plug it.
I will also let you name the algorithm, like that guy Nagle (he is in fact an a***hole given how much I have been strugling with it the last couple of days).
To expand a little on my strategy:
The way I incorporate upgrade time is in calculating the cost of the upgrade. So Total Upgrade Cost = upgrade cost + income lost during upgrade. Longer upgrades = greater income lost.
For me, ROI = (Total Upgrade Cost / Addtl IpH) + Upgrade Time
What I do is sort my next available upgrades by increase in IpH, then based on my cash on hand, I upgrade the highest increase in IpH where the ROI is not worse then any upgrade above it.
Example
Buidling - IpH increase - ROI
A - 10,000 - 100
B - 7,500 - 95
C - 6,000 - 110
D - 4,000 - 90
E - 3,000 - 95
F - 2,000 - 80
So, I would upgrade the one I can afford in this order (A,B,D,F). Eventually, my list looks something like this:
A - 10,000 - 100
B - 7,500 - 105
C - 6,000 - 110
D - 4,000 - 103
E - 3,000 - 115
F - 2,000 - 125
.....
Z - 100 - 80
In this case, I will upgrading Z and be forced to save for upgrade A.
I hope that makes sense.
Ram,
I am very interested to see how my strategy compares to GFI. I'm not sure which is better. I implicitly understand what I am trying to do. And why I am reluctant to use GFI again, is because I'm not convinced on the weighting you give to upgrade time. GFI may very well crush my strategy and if so, I will probably reluctantly convert to using it. My guess is that it's going to be very close and this discussion was all for naught.
Thanks again.
Just want to say as I converted the ROI inventor (Duder) to the Marginal Utility of Income method I use MUI which is a weighted index system, which incorporates upgrade(loss of income)time,cost/gain per dollar of construction/upgrade etc etc. I'm seeing this GFI (Gain For Investment) argument which I have to admit thought was gone way back in January 9 months ago! Again, build long, you're not aiming for a snapshot IPH of Y/Hr you are looking towards Level 10's. If you want to jump in puddles and pop into shops on the way to the harbour take the slow route. Casino's start well but appreciate on a slower dy/dx gradient than Nightclubs. You finish them, the rest fall into place. Hence a Level 10 NC is your first aim. Sharing upgrades between 2 means your jumping in puddles and doubling the time it takes to reach your ultimate Y/hr. I'll save for my Level 9 NC, $2.4bn and then Level 10 after that it just gets easier. I thought this was decided nearly a year ago but it still bumps along.....
Ramshutu
10-17-2012, 01:33 PM
What I do is sort my next available upgrades by increase in IpH, then based on my cash on hand, I upgrade the highest increase in IpH where the ROI is not worse then any upgrade above it.
I thought I understood what you are saying, but after starting some implementation, I have a bit of a problem.
If you sort by IPH after each upgrade, and choose the first item you can afford; there will NEVER be an item ABOVE the one you select that you can afford. Unless there is a "Save" rather than spend factor that you have not explained.
Hence a Level 10 NC is your first aim. Sharing upgrades between 2 means your jumping in puddles and doubling the time it takes to reach your ultimate Y/hr. I'll save for my Level 9 NC, $2.4bn and then Level 10 after that it just gets easier. I thought this was decided nearly a year ago but it still bumps along.....Wait, are you still advocating for going to a level 10 NC before building the second? I already debunked that myth. Check the sig. If you're saying something else, ignore me.
I thought I understood what you are saying, but after starting some implementation, I have a bit of a problem.
If you sort by IPH after each upgrade, and choose the first item you can afford; there will NEVER be an item ABOVE the one you select that you can afford. Unless there is a "Save" rather than spend factor that you have not explained.
I choose the first item I can afford AND doesn't have a worse ROI then anything above it. Eventually I keep exhausting mid-level better ROI investments and start to do cheaper upgrades, which forces me to save.
So in my example, using the first group, I skip C and E
Wait, are you still advocating for going to a level 10 NC before building the second? I already debunked that myth. Check the sig. If you're saying something else, ignore me.
We agree on this...
Ramshutu
10-17-2012, 01:52 PM
I choose the first item I can afford AND doesn't have a worse ROI then anything above it. Eventually I keep exhausting mid-level better ROI investments and start to do cheaper upgrades, which forces me to save.
So in my example, using the first group, I skip C and E
Do you search down the list, or up? If you are searching from high IPH to low this doesn't work. If you are searching from high to low, and chose the first item you can afford, by definition there will be no items above it that you can afford!
I search from highest IpH to low, then choose the first one I can afford AND has the lower (better) ROI then all upgrades above it.
Buidling - IpH increase - ROI
A - 10,000 - 100
B - 7,500 - 95
C - 6,000 - 110
D - 4,000 - 90
E - 3,000 - 95
F - 2,000 - 80
Decision tree looks something like this
- Can I afford A? no, move down list
- does B have better ROI then A? yes, can I afford it? no, move down list
- does C have better ROI then A and B? no, move down list
- does D have better ROI then A,B and C? yes; can I afford it? yes. do upgrade
I search from highest IpH to low, then choose the first one I can afford AND has the lower (better) ROI then all upgrades above it.
Buidling - IpH increase - ROI
A - 10,000 - 100
B - 7,500 - 95
C - 6,000 - 110
D - 4,000 - 90
E - 3,000 - 95
F - 2,000 - 80
Decision tree looks something like this
- Can I afford A? no, move down list
- does B have better ROI then A? yes, can I afford it? no, move down list
- does C have better ROI then A and B? no, move down list
- does D have better ROI then A,B and C? yes; can I afford it? yes. do upgrade
Eventually, my list looks something like this:
A - 10,000 - 100
B - 7,500 - 105
C - 6,000 - 110
D - 4,000 - 103
E - 3,000 - 115
F - 2,000 - 125
.....
Z - 100 - 80
In this case, if I can't afford A, Z is the next upgrade that has a better ROI then all above upgrades and therefore will then be forced to save for upgrade A.
Ramshutu
10-17-2012, 01:59 PM
Ahhhhhhh..... Gotcha.
Give me 15 minutes.
What name do you want to give it? :)
Euchred
10-17-2012, 02:01 PM
I don't mean to be ignorant but I kind of am. To me it seems crazy to ignore an NC upgrade. It's literally my only focus, any other upgrade I happen to complete at this point is just something to do while I amass funds... been working so far
Ramshutu
10-17-2012, 02:10 PM
Minor Coding error fixed.... Calculating.
mxz....your inbox in full...thanks for pm...but can't reply
Ahhhhhhh..... Gotcha.
Give me 15 minutes.
What name do you want to give it? :)
I'm not creative...can't think of a simple, descriptive name
I don't mean to be ignorant but I kind of am. To me it seems crazy to ignore an NC upgrade. It's literally my only focus, any other upgrade I happen to complete at this point is just something to do while I amass funds... been working so far
For the most part I think you are right...but I think that is far too simplistic, and may be missing some worthwhile upgrades, for those other top-end buildings (Palms, Casinos, Office, GC, etc)
montecore
10-17-2012, 02:15 PM
I don't mean to be ignorant but I kind of am. To me it seems crazy to ignore an NC upgrade. It's literally my only focus, any other upgrade I happen to complete at this point is just something to do while I amass funds... been working so far
That is basically mxz's premise and it's basically accurate. Once you have two level 10 night clubs, it really doesn't matter if your lofts, movie theaters, or anything else are level 10 or nonexistent.
I look for low ROI on buildings I like (12/24/48) and can see myself owning long term (income per square value at a high level). I tend to cap my upgrades at about 1/4th to 1/3rd of the total I'm saving up for, or else it's too easy to get tempted by 40, 50, 60 million upgrades while saving up for NC level 3 upgrades. And then you're just pushing a boulder up a hill to watch it roll back down again.
mxz....your inbox in full...thanks for pm...but can't replyGah, cleaned now.
We agree on this...We probably agree on a lot of things but can't get on the same page, haha.
Gah, cleaned now.We probably agree on a lot of things but can't get on the same page, haha.
Haha...yes, I think we do.
Ramshutu
10-17-2012, 02:24 PM
Here we go.... some results.
IPH Comparrison
http://i1063.photobucket.com/albums/t511/ramshutu/GraphIPH.jpg]
COH Comparrison
http://i1063.photobucket.com/albums/t511/ramshutu/GraphCOH.jpg
I'm actually suprised. The algorithm held up pretty well, it outperforms GFI for the first few $m/hr; but is generally the same, or underperforms against simply picking the highest IPH building you can afford.
So you're saying these discussions were pointless. Great. :)
Thanks for the sim, Rams, kind of cool to see it in graph form. Can you cut off around maybe 200 days? Or maybe only show days 100-300s to open things up a bit?
Ramshutu
10-17-2012, 02:29 PM
I have, however, spotted a bit of a problem that could skew results, I was strugglign to explain why you always had so much cash on hand. The reason that it assumes you can build all buildings.
I am going to do a quick tweak for the following:
1.) Allow all low level buildings.
2.) Only use buildings that are >= 6hr collect. This may have a difference by reducing overall available cash flow.
Ramshutu
10-17-2012, 02:30 PM
So you're saying these discussions were pointless. Great. :)
Thanks for the sim, Rams, kind of cool to see it in graph form. Can you cut off around maybe 200 days? Or maybe only show days 100-300s to open things up a bit?
Sure, that will be, tweaking a spin box! I accidently set the cut off at 1e7 rather than 1e6
Very amusing...so in end you are saying that it's almost makes no difference what strategy you implement. I do find that at least somewhat interesting.
I think what happens is that the Nightclub is such a dominant investment that no matter what strategy you implement, as long as it includes upgrading Nightclubs, then all the rest is basically noise.
Thank you very much Ram.
Ramshutu
10-17-2012, 02:40 PM
I re-ran with what I said, and I get similar answers; although Murf, your algorithm performs marginally better than GFI and IPH/H.
I need to work out how to tweak the methodology a little bit more to be more accurate. You can see by the COH graphs, that the bottleneck in the simulation is normally upgrade resources rather than cash available. I feel that in real situations, it is the cash on hand that you are always waiting for.
I re-ran with what I said, and I get similar answers; although Murf, your algorithm performs marginally better than GFI and IPH/H.
I need to work out how to tweak the methodology a little bit more to be more accurate. You can see by the COH graphs, that the bottleneck in the simulation is normally upgrade resources rather than cash available. I feel that in real situations, it is the cash on hand that you are always waiting for.
Yes, the COH chart seems a little odd. I would think it would reset closer to zero when each strategy starts a max upgrade it can afford. But since each strategy is experiencing similar oddity, fixing it should only tweak the results a little.
Thanks again for taking the time to do this. I think it was worth it, at least from my end as I didn't have to do all the work :).
prostouser
10-17-2012, 04:26 PM
Looks like there is some kind of agreement here? :)
I've compared 2 scenarios with figures from 1st post (accounting less IPH during upgrade, handwork only...)
Scenario 1: SaveForCasino3-UpgradeCasino3-SaveForNC7-UpgradeNC7
Scenario 2: SaveForNC7-UpgradeNC7-SaveForCasino3-UpgradeCasino3-Wait0.54Hr
Results after completing scenarios:
Scenario1: TotalTime = 450,2hr; CashOnHand = $255,9mln; FreeUpgradeTimer = 289,2hr
Scenario2: TotalTime = 450,2hr; CashOnHand = $254,7mln; FreeUpgradeTimer = 289,2hr
So Building NC7 after Casino3 is marginally better from point of COH after completing the task
BTW without Dices (faster upgrade ime) Scenario2 takes longer than scenario 1 with CoH1 = CoH2 + $5mln. It means that difference between scenarios are almost neglectable and will be much more heavily affected by other factors than order of upgrades (in real game).
...
To illustrate this, I plugged these two investments into a spreadsheet and calculated 200 days of these two investments.
I assumed we already had enough cash on hand for the Casino upgrade. I also assumed we kept collecting the $3M/hr. I also assumed we didn't lose the IPH during the upgrade (you'll see why in a minute this doesn't really matter).
So, on day 1, Scenario 1 started with a C3 upgrade and Scenario 2 started with a save for the NC 7.
...
As you can see, ROI is poor indicator of which is the better investment.
http://imageshack.us/a/img31/4936/c3nc7cash.jpg
...
Can I ask if it's not a secret: what tools you use to build this plots and how much time it took for this example?
Here we go.... some results.
IPH Comparrison
http://i1063.photobucket.com/albums/t511/ramshutu/GraphIPH.jpg]
COH Comparrison
http://i1063.photobucket.com/albums/t511/ramshutu/GraphCOH.jpg
I'm actually suprised. The algorithm held up pretty well, it outperforms GFI for the first few $m/hr; but is generally the same, or underperforms against simply picking the highest IPH building you can afford.
Just for curiosity - we all can guess that difference will be minimal - is it hard to model algorithm similar to MurfIPH but replacing "IPH increase" with "GFI" as an input parameter for decision making?
Buidling - GFI (instead of IpH increase) - ROI
A - 300 - 100
B - 250 - 95
C - 200 - 110
D - 150 - 90
E - 100 - 95
F - 50 - 80
Euchred
10-17-2012, 04:31 PM
For the most part I think you are right...but I think that is far too simplistic, and may be missing some worthwhile upgrades, for those other top-end buildings (Palms, Casinos, Office, GC, etc)
One does not make 5 million an hour by ignoring those buildings. I simply wait till they're made more affordable by upgrading NCs.
PawnXIIX
10-17-2012, 04:35 PM
What did I miss? ._.
Can I ask if it's not a secret: what tools you use to build this plots and how much time it took for this example?This one was Open Office, but Excel is pretty much the same. Maybe 15 minutes from scratch...on my other computer I have the templates ready and can compare upgrades in about 2-3 minutes.
PawnXIIX
10-17-2012, 04:44 PM
This one was Open Office, but Excel is pretty much the same. Maybe 15 minutes from scratch...on my other computer I have the templates ready and can compare upgrades in about 2-3 minutes.
Sorry to go off topic, but whenever I try to make a chart on OpenOffice it goes NOPE and crashes :(
One does not make 5 million an hour by ignoring those buildings. I simply wait till they're made more affordable by upgrading NCs.
Some people strictly do NC upgrades and nothing else, that is what I assumed your strategy was what you said what you did. Even with earning $5mm/hr or $100mm+/day, that still doesn't allow you to do $250mm upgrades to pass the time. These are somewhat meaningful upgrades that do slow your savings for your NC quite significantly. So, you are doing most of the upgrades that I would do, but just in a slightly different order.
Each of these strategies, yours, plux's, mine, Ram's, mzx's, they are very similar, and the best one will only be a small percentage better. And following the best strategy might get you to $5mm a few weeks ahead of the worst one. So, this is mostly a theoretical discussion and since this is really all I have left to discuss on the forums, because the rest has gone to crap, I choose to focus on it. Or else I wouldn't post anything.
Powered by vBulletin® Version 4.2.2 Copyright © 2024 vBulletin Solutions, Inc. All rights reserved.