Roulette strategy backtesting tool - Python

[meister245]

Hello everyone,

As a hobby, I look out for ways to make money online, and I started looking at roulette about 3-4 weeks ago,
In hindsight, I was extremely lucky to make 77 euros with an initial investment of 5 euros at the start, but then I lost it all 

So I've developed an open source "pattern based" roulette backtesting tool, to analyse strategies, if any of them can hold up in the long run. I got some good ideas from this forum, including where to look for historical wheel spin data and tips on roulette bot development, so I would give something back and (possibly) crowd source plausibly good ideas. I know about roulette as much as any other guy off the street.

I consider the tool mainly finished, I'm not planning on actively developing / expanding it, so you can hack it all you want. I would rather move on to developing other projects.

You can find the application here, all usage info is in the README
link:s:// gitlab.com/meister245/roulette-backtest (remove space)

As a demo, I was backtesting for rare patterns with a combination of martingale and I did find  a profitable pattern. As a disclaimer the test sample size is 1461 numbers per file and the occurrence is so rare, it's not feasible without a bot (+ probably they would kick your session for idling)

[Firefox]

Interesting, but since the wheel is an inanimate device is there a logical reason why past spins should influence coming ones. I'm talking in terms just of analysis of number patterns, not physical influences of wheel types, spin speeds or dealers.

[meister245]

Interesting, but since the wheel is an inanimate device is there a logical reason why past spins should influence coming ones. I'm talking in terms just of analysis of number patterns, not physical influences of wheel types, spin speeds or dealers.

This is just my personal view, with no statistics to back it up, just simple reasoning.

If we counted all the times on a sample set of 1000 spins, how many times we get streaks of 3 red numbers, we would get a much higher number, compared to if we counted all the instances of streaks of 6 reds happening. It would still hold true, that each number has the same probability of getting picked, however is this true for *streaks* of certain numbers as well?

If there are ~400 streaks of 3 red numbers and ~100 streaks of 6 red numbers getting picked, with this logic, I'm actually betting against long streaks. Like I'm betting that most of the time streaks will not last after reaching N length. The system is still fallible, it only depends how high you push the streak count.

Do correct me if I'm wrong, statistics was never my favorite.