Your test result on the same calculation method for martingale, in which after you lose, multiply by x. That's one of the way to do martingale and known to be failed early.
Check the following stat of my own martingale method (not publicity exist yet and no i don't publish it) and tell me how it can survive (21 million bets+).
https://imgur.com/a/1SIspbcIf you want to prove more, I will get a screenshot after it reach 100m bets, keep follow up since it's real betting thus taking like 10-11 months to have that.
You've earned 40$ after 21m bets while risking 750 $ if martingale will go too far? With regular martingale ~30% gamblers are lucky enough to double. You are far from doubling. Even if you will double your portfolio (earn 15 times more than you have already earned) you could be in those 30% lucky gamblers - it does not proof that your strategy is working. In fact if you will proof that it works you will open door to new mathematics in which 1+1=/2 because simple math proofs that it is impossible to beat casino.
You may mistaken from budget vs wagered amount, my budget is 1 LTC which is 66$ or at the time I input in @100$. With that 66$ and wagered 750$, it's performance is far better than the house edge, which is 40/750 = 5.3% earning per whatever wagered amount goes, and the drawdown is not even reaching 50% of it (aka 33$). So current capital is 66$+40$ = 104$.
But you have based on dollar cost averaging thus not applicable in this case, since the site translate everything into BTC for stat and the real stat is: 1LTC budget, 0.72876726 LTC profit after 8.83229109 LTC wagered or 72% profit per budgeted amount, or 8.25% per wagered amount. The key is the later percentage in which it keep following the wagered amount (not capital) the more it wager, the more it earn and as long as this percentage is greater than the house edge of 1%, it alive as long as it should be because it is harder to lose since the profit increase, and the running balance keep increase as well as drawdown is less than 50%.
Majority of the tests out there do not take into account the earning rate (running current balance) but only calcualate base on original balance. They missed the "moving variable" which is the earning (or profit), as long as it go faster than what is required to win a rare streak, it's still success.
For test, yes I did test but simulated of each win/lose bet including the earning too which is "bottom up" to get the final summarization, not by simple formula base on few variables which is "generic" or "top down". If you ever hear of bottom up or top down estimation/calculation, you will know that bottom up method yeid a more accurate result than the other. And the number of test I've done (which is the most important factor that affect the result) is 200 millions at least, as well as time taken to do so is few weeks, not by a simply click where the computer give the result immediately.