Here I do a recap of my sets' historical performances, as well as those of a couple other popular bettors. I use my statistical models, as well as the actual closing odds of the pirates every day, in order to simulate every day's results 1 million times, and see how each different strategy performs in the long term in all those simulations. The advantage of this simulation over simply tracking ROIs is that it shows us not only how much each strategy outperformed or underperformed expectations, but also exactly how much luck was involved, and to what degree that luck impacted the performance.
For each strategy/bettor, I illustrate the results in three different ways. After all the individual strategy visualizations, I also do some distribution comparisons between similar strategies. Check here for instructions on how to read the data.
To be clear, we only know all of these numbers in retrospect. It is therefore not correct to assume that if you follow any of my sets or any other bettors' sets for the next month, that your expected profit will continue to have the exact likelihoods listed. This is especially true because every day's odds are different, and some months are simply better or worse than other ones. However, I think this is still a useful illustration of the distributions of different strategies, and of what you can generally expect when following a particular strategy.
Nsheng Beginner 
| Units of profit: | 8733 |
| ROI: | 1.64 |
| Expected units of profit: | 8844 |
| Expected ROI: | 1.65 |
| Luck%: | 34% |
| Percentile | Units of Profit | ROI |
|---|
| 5th Percentile | 8405 units | 1.62 |
| 25th Percentile | 8665 units | 1.64 |
| 34th Percentile | 8733 units | 1.64 |
| 50th Percentile | 8845 units | 1.65 |
| 75th Percentile | 9024 units | 1.67 |
| 95th Percentile | 9279 units | 1.68 |
| ROI Range | Probability | CProb | Tail |
|---|
| Lower than 1.61 ROI | 1.5% | 1.5% | 100.0% |
| Between 1.61 and 1.62 ROI | 3.4% | 4.9% | 98.5% |
| Between 1.62 and 1.63 ROI | 7.5% | 12.4% | 95.1% |
| Between 1.63 and 1.64 ROI | 13.4% | 25.8% | 87.6% |
| Between 1.64 and 1.65 ROI | 18.5% | 44.3% | 74.2% |
| Between 1.65 and 1.66 ROI | 20.1% | 64.4% | 55.7% |
| Between 1.66 and 1.67 ROI | 16.7% | 81.1% | 35.6% |
| Between 1.67 and 1.68 ROI | 10.8% | 91.9% | 18.9% |
| Between 1.68 and 1.69 ROI | 5.3% | 97.3% | 8.1% |
| Higher than 1.69 ROI | 2.7% | 100.0% | 2.7% |
Nsheng Standard
| Units of profit: | 10968 |
| ROI: | 1.81 |
| Expected units of profit: | 11522 |
| Expected ROI: | 1.85 |
| Luck%: | 13% |
| Percentile | Units of Profit | ROI |
|---|
| 5th Percentile | 10720 units | 1.79 |
| 13th Percentile | 10968 units | 1.81 |
| 25th Percentile | 11192 units | 1.83 |
| 50th Percentile | 11522 units | 1.85 |
| 75th Percentile | 11852 units | 1.87 |
| 95th Percentile | 12324 units | 1.91 |
| ROI Range | Probability | CProb | Tail |
|---|
| Lower than 1.78 ROI | 2.5% | 2.5% | 100.0% |
| Between 1.78 and 1.80 ROI | 5.6% | 8.1% | 97.5% |
| Between 1.80 and 1.82 ROI | 11.9% | 20.0% | 91.9% |
| Between 1.82 and 1.84 ROI | 18.7% | 38.8% | 80.0% |
| Between 1.84 and 1.86 ROI | 21.8% | 60.6% | 61.2% |
| Between 1.86 and 1.88 ROI | 18.9% | 79.5% | 39.4% |
| Between 1.88 and 1.90 ROI | 12.1% | 91.6% | 20.5% |
| Between 1.90 and 1.92 ROI | 5.8% | 97.4% | 8.4% |
| Higher than 1.92 ROI | 2.6% | 100.0% | 2.6% |
Nsheng Aggressive
| Units of profit: | 11403 |
| ROI: | 1.84 |
| Expected units of profit: | 12063 |
| Expected ROI: | 1.89 |
| Luck%: | 15% |
| Percentile | Units of Profit | ROI |
|---|
| 5th Percentile | 11007 units | 1.81 |
| 15th Percentile | 11403 units | 1.84 |
| 25th Percentile | 11627 units | 1.86 |
| 50th Percentile | 12062 units | 1.89 |
| 75th Percentile | 12498 units | 1.92 |
| 95th Percentile | 13126 units | 1.97 |
| ROI Range | Probability | CProb | Tail |
|---|
| Lower than 1.80 ROI | 2.8% | 2.8% | 100.0% |
| Between 1.80 and 1.85 ROI | 17.1% | 19.9% | 97.2% |
| Between 1.85 and 1.90 ROI | 38.3% | 58.2% | 80.1% |
| Between 1.90 and 1.95 ROI | 31.3% | 89.5% | 41.8% |
| Between 1.95 and 2.00 ROI | 9.5% | 99.0% | 10.5% |
| Higher than 2.00 ROI | 1.0% | 100.0% | 1.0% |
Nsheng Adventurous
| Units of profit: | 11418 |
| ROI: | 1.84 |
| Expected units of profit: | 12450 |
| Expected ROI: | 1.92 |
| Luck%: | 13% |
| Percentile | Units of Profit | ROI |
|---|
| 5th Percentile | 10957 units | 1.81 |
| 13th Percentile | 11418 units | 1.84 |
| 25th Percentile | 11826 units | 1.87 |
| 50th Percentile | 12442 units | 1.92 |
| 75th Percentile | 13066 units | 1.96 |
| 95th Percentile | 13970 units | 2.03 |
| ROI Range | Probability | CProb | Tail |
|---|
| Lower than 1.80 ROI | 3.8% | 3.8% | 100.0% |
| Between 1.80 and 1.85 ROI | 11.7% | 15.5% | 96.2% |
| Between 1.85 and 1.90 ROI | 23.9% | 39.4% | 84.5% |
| Between 1.90 and 1.95 ROI | 28.5% | 67.9% | 60.6% |
| Between 1.95 and 2.00 ROI | 20.5% | 88.4% | 32.1% |
| Between 2.00 and 2.05 ROI | 8.9% | 97.3% | 11.6% |
| Higher than 2.05 ROI | 2.7% | 100.0% | 2.7% |
| Units of profit: | 11943 |
| ROI: | 1.88 |
| Expected units of profit: | 11963 |
| Expected ROI: | 1.88 |
| Luck%: | 49% |
| Percentile | Units of Profit | ROI |
|---|
| 5th Percentile | 10893 units | 1.80 |
| 25th Percentile | 11518 units | 1.85 |
| 49th Percentile | 11943 units | 1.88 |
| 50th Percentile | 11959 units | 1.88 |
| 75th Percentile | 12404 units | 1.92 |
| 95th Percentile | 13046 units | 1.96 |
| ROI Range | Probability | CProb | Tail |
|---|
| Lower than 1.80 ROI | 4.2% | 4.2% | 100.0% |
| Between 1.80 and 1.85 ROI | 20.8% | 25.0% | 95.8% |
| Between 1.85 and 1.90 ROI | 39.1% | 64.0% | 75.0% |
| Between 1.90 and 1.95 ROI | 27.6% | 91.6% | 36.0% |
| Higher than 1.95 ROI | 8.4% | 100.0% | 8.4% |
Garet 
| Units of profit: | 11140 |
| ROI: | 1.82 |
| Expected units of profit: | 11036 |
| Expected ROI: | 1.81 |
| Luck%: | 57% |
| Percentile | Units of Profit | ROI |
|---|
| 5th Percentile | 10086 units | 1.74 |
| 25th Percentile | 10642 units | 1.79 |
| 50th Percentile | 11033 units | 1.81 |
| 57th Percentile | 11140 units | 1.82 |
| 75th Percentile | 11426 units | 1.84 |
| 95th Percentile | 11994 units | 1.89 |
| ROI Range | Probability | CProb | Tail |
|---|
| Lower than 1.75 ROI | 6.6% | 6.6% | 100.0% |
| Between 1.75 and 1.80 ROI | 30.5% | 37.0% | 93.4% |
| Between 1.80 and 1.85 ROI | 42.6% | 79.7% | 63.0% |
| Between 1.85 and 1.90 ROI | 18.0% | 97.7% | 20.3% |
| Higher than 1.90 ROI | 2.3% | 100.0% | 2.3% |
| Units of profit: | 10818 |
| ROI: | 1.80 |
| Expected units of profit: | 9062 |
| Expected ROI: | 1.67 |
| Luck%: | 90% |
| Percentile | Units of Profit | ROI |
|---|
| 5th Percentile | 6919 units | 1.51 |
| 25th Percentile | 8151 units | 1.60 |
| 50th Percentile | 9038 units | 1.67 |
| 75th Percentile | 9948 units | 1.73 |
| 90th Percentile | 10818 units | 1.80 |
| 95th Percentile | 11287 units | 1.83 |
| ROI Range | Probability | CProb | Tail |
|---|
| Lower than 1.50 ROI | 3.9% | 3.9% | 100.0% |
| Between 1.50 and 1.55 ROI | 7.3% | 11.2% | 96.1% |
| Between 1.55 and 1.60 ROI | 13.4% | 24.5% | 88.8% |
| Between 1.60 and 1.65 ROI | 18.6% | 43.1% | 75.5% |
| Between 1.65 and 1.70 ROI | 20.0% | 63.1% | 56.9% |
| Between 1.70 and 1.75 ROI | 16.7% | 79.7% | 36.9% |
| Between 1.75 and 1.80 ROI | 11.0% | 90.7% | 20.3% |
| Between 1.80 and 1.85 ROI | 5.8% | 96.5% | 9.3% |
| Between 1.85 and 1.90 ROI | 2.5% | 99.0% | 3.5% |
| Higher than 1.90 ROI | 1.0% | 100.0% | 1.0% |
Distribution Comparisons
Profit Comparisons
How to read the data
Summary
- Units of profit: Number of units won minus number of units played.
- ROI: Units of profit divided by (10 times number of days in the simulation). Note that for this calculation of ROI, skipping is equivalent to breaking even.
- Expected units of profit: Number of units expected to profit on average.
- Expected ROI: Expected ROI on average.
- Luck%: How lucky this strategy's actual results were. This is the percentile of the strategy's actual profit, amongst all the possible results in the simulation. It is a number between 0% and 99%, inclusive.
I want to underscore here the usefulness of the Luck% metric. We often say that a bettor got lucky or unlucky in a particular month. Not only does Luck% assign a numerical value to how lucky they got, but more importantly, this number accounts for how risky the bettor's strategy is. If a bettor has an average TER of 15 but engages in extremely risky strategies, it doesn't take too much luck for them to hit 2.5 ROI for the month, nor too much un-luck to hit 0.5 ROI either. So for example, 2.5 ROI might represent 80% luck for that bettor, while 0.5 ROI might represent 20% luck. Meanwhile, a safe bettor with an average TER of 15 will require an impossibly large amount of luck to hit 2.5 ROI, and similarly an impossibly large amount of un-luck to hit 0.5 ROI. So for the safe bettor, those might represent 99% and 0% luck. In this way, Luck% gives us an objective answer to the question How lucky was this strategy?, regardless of how safe or how crazy the strategy is.
Histogram with box-and-whisker plot
The histogram shows us how often each outcome occurred in the simulations. The x-axis shows the number of units of profit from the month, and the y-axis indicates frequency.
The box-and-whisker plot shows major percentiles in the data:
- The left whisker indicates the 5th percentile. So, 5% of simulations resulted in profits to the left of there.
- The left side of the box indicates the 25th percentile. So, 25% of simulations resulted in profits to the left of there.
- The line in the middle of the box indicates the 50th percentile, also called the median.
- The green triangle indicates the average. For safe strategies, the average will be virtually identical to the median.
- The right side of the box indicates the 75th percentile.
- The right whisker indicates the 95th percentile.
The magenta line indicates the result that actually happened. The red curve fits an approximate distribution to the data.
What questions can this plot answer?
- Where are the most likely results concentrated?
Check the values where the histogram is at the highest!
- How wide is the range of likely outcomes?
Check how wide the box and its whiskers are!
- How lucky was this strategy?
Check how far to the left or right of the histogram the magenta line is. The further to the right, the luckier the strategy was!
Percentiles table
| Percentile | Units of Profit | ROI |
|---|
| 5th Percentile | 122 units | 1.39 |
| 18th Percentile | 204 units | 1.66 |
| 25th Percentile | 226 units | 1.73 |
| 50th Percentile | 304 units | 1.98 |
| 75th Percentile | 384 units | 2.24 |
| 95th Percentile | 503 units | 2.62 |
In this table, I simply list the data points from the box-and-whisker plot above, as well as the actual percentile that this strategy ended up landing in for the duration simulated, highlighted in magenta.
What questions can this table answer?
- If I had followed this strategy, and had completely average luck, how much profit would I have gotten?
Check the "50th percentile" row!
- What about if I had good luck?
Check "75th percentile"!
- What about absolutely horrible luck?
Check "5th percentile"!
- How lucky did this strategy actually get?
Check the percentile of the magenta row! This is the same number as the Luck% from the summary.
ROI table
| ROI Range | Probability | CProb | Tail |
|---|
| Lower than 1.0 ROI | 1.6% | 1.6% | 100% |
| Between 1.0 and 1.2 ROI | 3.3% | 4.9% | 98.4% |
| Between 1.2 and 1.4 ROI | 6.4% | 11.3% | 95.1% |
| Between 1.4 and 1.6 ROI | 10.2% | 21.5% | 88.7% |
| Between 1.6 and 1.8 ROI | 13.1% | 34.6% | 78.5% |
| Between 1.8 and 2.0 ROI | 14.7% | 49.3% | 65.4% |
| Between 2.0 and 2.2 ROI | 14.2% | 63.5% | 50.7% |
| Between 2.2 and 2.4 ROI | 12.1% | 75.6% | 36.5% |
| Between 2.4 and 2.6 ROI | 9.2% | 84.8% | 24.4% |
| Between 2.6 and 2.8 ROI | 6.5% | 91.3% | 15.2% |
| Between 2.8 and 3.0 ROI | 4.0% | 95.3% | 8.7% |
| Higher than 3.0 ROI | 4.7% | 100% | 4.7% |
This view of the data tells you how likely each ROI band was, and is purposely organized so as to be familiar to bettors who are accustomed to reading the NFC Odds table. I also highlight the ROI band that this strategy actually fell in for the duration simulated.
- ROI Range - The first column is a range of possible ROI results.
- Probability - The second column indicates the probability of the strategy hitting an ROI within that range for the duration simulated.
- CProb - The third column indicates the probability of the strategy hitting an ROI within that range, or lower.
- Tail - The fourth column indicates the probability of the strategy hitting an ROI within that range, or higher.
What questions can this table answer?
- What's the chance that I at least double the NP I put in?
Check the Tail column of the Between 2.0 and 2.2 row!
- What's the chance that I don't even break even on the NP I put in?
Check the Probability column of the Lower than 1.0 ROI row!
Distribution comparisons
This graph displays fitted distribution curves for a few different strategies. The x- and y-axes have the same meaning as in the histogram.
What questions can this graph answer?
- Which of the strategies is safer?
In general, for two "bell-shaped" distributions, the taller distribution is safer, because the likely results are more concentrated in a small range near the center. For example, in this graph, the green strategy is safer. Another way you can look at this is, the taller distribution is narrower -- it doesn't flair out as wide on the left and right. That means that even if terrible luck strikes and you end up all the way on the left of the distribution, it still won't be that bad. Compare that to the wider red distribution, which has a higher chance to hit a really terrible ROI, if it gets very unlucky.
- Which of the strategies has generally higher payouts?
In general, for two "bell-shaped" distributions, the distribution further to the right has generally higher payouts, because results to the right are higher. For example, in this graph, the red strategy has generally higher payouts.
- Which of the strategies has a higher chance to hit around 300 units of profit for the duration simulated?
Check which distribution is higher above the 300 mark on the x-axis! For example, in this graph, the green strategy has a higher chance to hit around 300 units of profit.
- Which of the strategies is the best/strongest?
That's up to you to decide! Usually higher payout strategies will also be riskier, so you have to figure out what the right balance is for you. Also, remember that at the end of the day, luck still plays a large factor in this game. Even if you pick a strategy that is better or safer than another strategy, the other strategy may simply get luckier than the one you picked.
Profit comparisons
All of the above results and visualizations are given in units, because it's easier to compute, and doesn't require us to factor in your personal max bet, based on your account age. However, units are very abstract. In reality, what we really care about is how many Neopoints we win. These four graphs tell us:
- Actual Neopoints of Profit: How many Neopoints did each strategy actually profit, given your current max bet? For this calculation, we also include Neopoints saved when you don't make your full max bet due to the 1 million NP ceiling.
- Actual ROI: Neopoints of Profit divided by the total number of Neopoints you were allowed to bet, given your current max bet.
- Expected Neopoints of Profit: How many Neopoints of profit did each strategy expect to make, given your current max bet?
- Expected ROI: Expected Neopoints of Profit divided by the total number of Neopoints you were allowed to bet, given your current max bet.
For simplicity, I calculate bet amounts for each bet using the round's closing odds. This is not a very realistic assumption, especially for the risky sets, but should get us close enough to actual performance numbers. For comparison, I also include a dotted line for bet amounts computed using opening odds.
What questions can these graphs answer?
- If I had followed the red strategy, how many Neopoints would I have profited during this time?
On the "Neopoints of Profit" graph, find your current personal max bet amount on the x-axis, and check the y value of the red line there! For example, in this graph, if your current max bet is 10,000 NP, then the red strategy would have gotten you about 20,000 NP of profit during this time.
- Which strategy will win me more Neopoints on average?
On the "Normalized Expected ROIs" graph, find your current personal max bet amount on the x-axis, and check which line is the highest there! For example, in this graph, if your max bet is less than 8,000 NP, then the green strategy will win you more Neopoints on average.
Comments
Post a Comment