Bankroll Requirements and Staking Plan

We have now arrived to the point we are confident with our probability calculations and understand value betting, and we want with no further delay to submit our bets and start making money. However, if we don’t bet according to a staking plan and not having defined any bankroll for our system, we will be lucky not to end up bankrupt. But what are the staking plans we can choose from, which one is most appropriate for our system and how can we calculate the bankroll we are going to need?

All these are found in this article being the fourth one on the Series about how to build your own Betting System.

Let’s start with the staking plans:

The first and most used plan is the flat betting plan. It is exactly what the name implies, betting with the same fixed amount on each wager. It doesn’t take into account any bankroll and you don’t have to calculate anything. It is the easiest method to apply. Still, I advise using it for the first 1,000 or so bets, because this plan can be used in order to see if the betting system is a winner and some other significant statistics of it. More on than later in this post.
Then comes the staking plan based on the Kelly criterion. I won’t get into details, there is adequate information and documentary in the internet (check here (wikipedia), or here), but I have to say that there are some advantages and disadvantages about it. There are also variations and modifications of it (check here and here(pdf)) and you are free to apply any of them. The main concept in simple words is that your bankroll will increase much faster with it (in case the betting system wins) but at the same time there will be wilder variance (swings) of the bankroll.
There are other staking plans (martingale, anti-martingale, etc.) but I won’t go deep into them. I have found the aforementioned plans to be the best regarding betting (and investing in general), at least during the first steps. If one has become an expert in the field, he may test or even invent his own sophisticated plan.

Regarding bankroll requirements, there have been a lot of discussions in forums and websites. I won’t bother you with the same stuff and I will try to get directly into the point. Given that we apply the flat staking plan, we will need 500-1,000 games in order to exact some useful conclusions. While we collect this sample data, we may in fact place the bets but I’ll suggest using very small stakes, the smallest available. By betting say €4 on each event, we would have staked €4,000 in the end. If our betting strategy loses, we would expect about 10-15% losses in the worst scenario. Our funds would have depleted around €400 after 1,000 bets. While it is still a big amount, I can fully understand the desire to apply the method as soon as possible. Just have in mind that it can be expensive.

Having collected the sample data, so that we are now ready to tune our system, there are some interesting statistics calculations we can do, in order to see if we have a winning strategy. While we are collecting sample data, it is very important to record the odds for all the possibilities of the events. the final result and whether our bet won or lost and how much. By doing this, we will be able to fine-tune later our system. But for now we will just need the amount that won or lost for each bet.

We calculate the average number of all the results (wins AND losses). If it is a positive number we would have won, if it a negative we would have lost. Note that if a row didn’t include an actual bet, that shouldn’t be added as zero but rather not included at all in the calculation. Don’t forget that (hint: write an “if” statement in excel defining a cell as empty (“”) if there was no betting in that row).
We calculate the standard deviation of those results. The formula is included in Excel (stdev).
Finally we calculate the confidence of our results. The formula is once again included (confidence) and we will use 30% significance level (alpha = 0.7).
Comparing this number with the average (mean) number, we want to see a difference of less than the average number. That is if the average is €1, we’d expect the confidence at 30% significance level to range between €0.10 to 0.90. That would mean we most probably have a winning strategy. If the number is bigger than the average (>1) then we can’t say for sure yet. You can read the formula’s help page found in Excel.

Since by now you’ll know if your betting system is showing steady profits, you may use the kelly criterion to calculate your stakes. Theoretically, using that criterion, you won’t experience bankruptcy ever in your betting career since the bets are always defined proportionally to the bankroll. For a flat betting plan I’d advise having a bankroll of at least 50 or 100 (better) bets. Finally I wouldn’t risk more than 2% of my bankroll on any bet, regardless of what my staking plan implies. A drawdown of 50 losing bets would have my bankroll erased.

Let’s see where we stand now: