Probability Statistic Confirmation

[Chippy]

I'm sorry to resurrect this thread, but I see this question come up in several other posts and didn't want to re-ask it. I've been playing around with various strategies and have one that indicates 5 trades executing... all 5 hit the profit target. The profit win average is exactly the same as the overall average. Yet the probability is 0. I'm having a hard time understanding what relevance "probability" has. The formula isn't published on the help page (not that I could find). Are we sure that it's working as intended?

[NinjaTrader_ChrisL]

Hi, thanks for posting.

You can plug your results into the students t-distribution to find out why this is producing 0, the method will return a 1 for true or 0 for false.
I found a community members re-implementation:

double div = systemPerformance.AllTrades.TradesPerformance.Percent.StdDev / Math.Sqrt(systemPerformance.AllTrades.TradesCount) ;
double t = Stat.StudTp(systemPerformance.AllTrades.TradesPerf ormance.Percent.AvgProfit / div, systemPerformance.AllTrades.TradesCount - 1);
return (div <= 0.5 ? 1 - t : t);

Kind regards,
-ChrisL

[Chippy]

I appreciate the response, but I think you have a bug here in the SystemPerformance.AllTrades.TradesPerformance.Perc ent collection.

btw... the "s" should be capitalized in "SystemPerformance."

I have run a Print() on just about all the properties in the Percent collection. The numbers returned do not represent anything even close to what they should. I ran this in a backtest strategy that had exactly six trades. Each of them had a take profit of $25. There were no losers. ie, Percent.AverageProfit returns a value of "0.00112206428499886" following the last trade in the dataset. So, if "AverageProfit" "LargestWinner" "LargestLoser" are all returning very similar values to 0.00112206428499886, I didn't bother messing with the rest of the method you posted, nor try to check NT's math on the StdDev.

[Chippy]

I just ran ...Percent.StdDev as well... while the number was small (on the order of 3 x 10^-6) it still was not zero... which is what the std deviation should be for a dataset where all data points are equal (6 trades, $25 each).

[NinjaTrader_ChelseaB]

Hello Chippy,

I will look into these metrics and I will let you know what I find.

I appreciate your patience.

[NinjaTrader_ChelseaB]

Hello Chippy,

I've made some output to see if this is calculating correctly. From what I am seeing the numbers are matching.

Note, that with percent metrics the profit is re-invested into the next trade and is cumulative.

A sample of the output from the last trade:

AllTrades.Count: 6, Currency.StdDev: 43.3304118940428, Percent.StdDev: 0.000197012793405613
2/14/2022 4:20:00 PM | trade.Quantity: 1, trade.ProfitCurrency -37.5, trade.ProfitPercent: -0.000170415814587594
(profitCurrency: -37.5 / quantity: 1 - currencySumProfit: 162.5 / TradesCount: 6): -64.5833333333333 * (profitCurrency / quantity - currencySumProfit / TradesCount) = 4171.00694444444
-
(profitPercent: -0.000170415814587594 - percentSumProfit: 0.000738930258560834 / TradesCount: 6): -0.000293570857681066 * (profitPercent - percentSumProfit / TradesCount) = 8.61838484795968E-08
√(currencyDevProfit: 11265.1475694444 / TradesCount): 1877.52459490741 = 43.3304118940428
√(percentDevProfit: 2.32884244592896E-07 / TradesCount): 3.88140407654827E-08 = 0.000197012793405613

Attached is the full output from 6 trades in the test.

[HelenaMaria]

The code is working perfectly.

[NinjaTrader_ChelseaB]

Hello HelenaMaria,

Welcome to the NinjaTrader forums!