Probability theory for trading signals
There is a standard mathematical task. Let’s assume that six guns fired simultaneously at one target.
The probability of hitting one target with one shot from one gun is 0.35 (35%); for the rest of the guns separately, the probability of hitting the target is also 0.35 (35%).
We need to find the probability of at least one hit in one salvo from all guns.
I believe that solution of this task can be used to find the correct moment for buy or sell signals. Let's say we have six indicators of different natures (stochastic, trend, chart pattern, volume profile, Fibonacci levels, candlestick patterns etc.) and each indicator has a relatively low win rate %, let's assume conservatively, that each indicator has 35% of win rate. If all indicators simultaneously show a buy signal, then the probability that at least one signal will be correct is calculated as follows:
The probability that each signal separately is true:
p1 = 35%, p2 = 35%, p3 = 35%, p4 = 35%, p5 = 35%, p6 = 35%
The probability that each signal separately is false:
q1 = 65%, q2 = 65%, q3 = 65%, q4 = 65%, q5 = 65%, q6 = 65%
The probability that all signals will be false at the same time:
P_(A) = q1 x q2 x q3 x q4 x q5 x q6 = 8%
The probability that at least one signal will be true:
P(A) = 1 - P_(A) = 1 – 0,08 = 0,92 (92%)
So, we just confirmed obvious thing that if many indicators show the same signal simultaneously that we have higher chance to win as compared with signal of only one technical indicator. But you must be careful.
- This methodology doesn’t work good in a flat market
- Big players can easily steal your stop-losses because they also see these cumulative signals.
- Close to 90% win-rate is not 100% win-rate, thus risk management is a must anyway
- this method is quite heavy due to big volume of manual work.
Important note: in order to avoid receiving too optimistic results, it’s necessary to use different types of indicators by their nature.
Don’t use, for example three stochastics and five trend indicators. Correct example of indicators: 1 Fibonacci levels, 1 stochastics, 1 candlestick pattern, 1 chart pattern, 1 volume profile, 1 trend indicator.
Anyway, I just made a test of this methodology for Crypto, Equity and Forex and the win-rate was about 80-90% with 80% profit after three weeks, I think that it really works.
The probability of hitting one target with one shot from one gun is 0.35 (35%); for the rest of the guns separately, the probability of hitting the target is also 0.35 (35%).
We need to find the probability of at least one hit in one salvo from all guns.
I believe that solution of this task can be used to find the correct moment for buy or sell signals. Let's say we have six indicators of different natures (stochastic, trend, chart pattern, volume profile, Fibonacci levels, candlestick patterns etc.) and each indicator has a relatively low win rate %, let's assume conservatively, that each indicator has 35% of win rate. If all indicators simultaneously show a buy signal, then the probability that at least one signal will be correct is calculated as follows:
The probability that each signal separately is true:
p1 = 35%, p2 = 35%, p3 = 35%, p4 = 35%, p5 = 35%, p6 = 35%
The probability that each signal separately is false:
q1 = 65%, q2 = 65%, q3 = 65%, q4 = 65%, q5 = 65%, q6 = 65%
The probability that all signals will be false at the same time:
P_(A) = q1 x q2 x q3 x q4 x q5 x q6 = 8%
The probability that at least one signal will be true:
P(A) = 1 - P_(A) = 1 – 0,08 = 0,92 (92%)
So, we just confirmed obvious thing that if many indicators show the same signal simultaneously that we have higher chance to win as compared with signal of only one technical indicator. But you must be careful.
- This methodology doesn’t work good in a flat market
- Big players can easily steal your stop-losses because they also see these cumulative signals.
- Close to 90% win-rate is not 100% win-rate, thus risk management is a must anyway
- this method is quite heavy due to big volume of manual work.
Important note: in order to avoid receiving too optimistic results, it’s necessary to use different types of indicators by their nature.
Don’t use, for example three stochastics and five trend indicators. Correct example of indicators: 1 Fibonacci levels, 1 stochastics, 1 candlestick pattern, 1 chart pattern, 1 volume profile, 1 trend indicator.
Anyway, I just made a test of this methodology for Crypto, Equity and Forex and the win-rate was about 80-90% with 80% profit after three weeks, I think that it really works.
Comment:
So, my educational setup has got +30% profit.