Sharp Ratio Calculation

[berndhart]

In the docs (https://ninjatrader.com/de/support/h...efinitions.htm), the Sharp Ratio is defined as:
(Profit per Month – risk free Rate of Return) / standard deviation of monthly profits

So my first question would be what if my range of data is smaller than a month? It seems like I still get a sharp value even though the standard deviation could only get 1 data point, in which case it couldn't be calculated.

My second question is regarding the calculation. I wanted to try it myself and I used the data which the Example Moving Average Strategy that is built in provides.
when using the "Net Profit"-values of each Month and and the Profit per Month result I get

(using python):
-1082.57 / np.std([1475.46, 3855.48, 1337.71, 3464.98, -10528.74, -6064.82]) = -0.20247716864741613

the Sharp Ratio which the Strategy Analyzer prints is -0.40. Why are they not the same, I used the calculation from the docs?
What am I missing here?

Screenshot with all the results + calculation:

[NinjaTrader_ChelseaB]

Hello berndhart,

While I cannot provide the direct code, I will work on creating output from prints to help you track this, similar to the post linked below.

https://ninjatrader.com/support/forum/forum/ninjatrader-8/add-on-development/1175134-understanding-ulcer-index?p=1177259#post1177259

In the meantime, you can retrieve this value from the SystemPerformance.AllTrades.TradesPerformance.Shar pRatio and print this after each position close or change to watch the progression.

https://ninjatrader.com/support/helpGuides/nt8/sharperatio.htm

[NinjaTrader_ChelseaB]

Hello berndhart,

I've gotten that script together to provide you some output to help you track how the SharpeRatio is calculated.

Last trade:

totalDays: 16, monthsNum: 0.524590163934426
(1.0 + percentCumProfit: -0.0193012326463206): 0.980698767353679 * (1.0 + profitPercent: -0.00111660551922157): 0.998883394480778 = -0.0203962863026415 + 1 = -0.0193012326463206
(|percentCumProfit: -0.0193012326463206 + 1| = 0.980698767353679 ^ 1.0 / monthsNum: 0.963529007002894 = -0.0364709929971057) - 1
(percentProfitPerMonth: -0.0364709929971057 - riskFreeReturn: 0): -0.0364709929971057 / monthlyStdDev: 0.0640149466792037 = -0.56972621065939

[berndhart]

Thank you for the answer,
I tried to understand your calculation, that is what I understood that far:
from the docs (https://ninjatrader.com/de/support/h...efinitions.htm) you used:

In the first line:

(1.0 + percentCumProfit: -0.0743994238517707): 0.925600576148229 * (1.0 + profitPercent: 0): 1 = -0.0743994238517707 + 1 = -0.0743994238517707

you are calculating:

SUM((1 + Current Profit in Percent) * (1 + Profit in Percent) - 1) for all trades

but for one trade only hence "Profit in Percent" = 0, since there weren't any trades before that point in time.

The in the next line:

(|percentCumProfit: -0.0743994238517707 + 1| = 0.925600576148229 ^ 1.0 / monthsNum: 0.952061506373716 = -0.0479384936262837) - 1

you are calculating:

(1 + cumulative profit)(1 * (30.5 / # days)) - 1

which gives us the Profit per Month in percent: -0.0479384936262837.

Now we get to the Sharpe Ratio, this is where I can't follow you, definition from docs:

(Profit per Month – risk free Rate of Return) / standard deviation of monthly profits

your calculation:

(currencyProfitPerMonth: -10635.2864583333 - riskFreeReturn: 0): -10635.2864583333 / monthlyStdDev: 0.0547242922537766 = -0.876000248737348

With the "currency profit per Month" being calculated as a ES contract * "Profit per Month in percent".
The "monthlyStdDev" is where I am still not understanding how it is calculated. Since the StdDev always needs more than 1 value to be calculated I don't understand how we can get a value from it with only 1 trade done to that point in time, also why is it 0.0547242922537766?

[NinjaTrader_Jim]

Hello berndhart,

Thanks for pointing out. Our test script was printing Currency parts of the calculation. I made corrections so it prints Percent information for the Percentage calculation.

I changed the information from post #3 to reflect the percentage calculation with fewer trades.

[Leeroy_Jenkins]

•NinjaTrader hard sets "risk-free Rate of Return" to a value of zero

Don't know why you guys did this but this leads to misleading results. It's not real Sortino or Sharpe ratios.

[NinjaTrader_Jim]

Hello Leeroy_Jenkins,

There wasn't a good way to get accurate data for RiskFreeReturn for all setups, which is why we have a RiskFreeReturn property for TradesPerformance where it can be overridden.

https://ninjatrader.com/support/help...freereturn.htm