Tweet
*** see attached PDF for correct tabulations ***
In the Back Test Report, I believe the "Monthly Returns & Drawdowns (%)" and the "Annual Returns & Drawdowns (%)" are being incorrectly calculated.
In short, the denominator used in the performance calculation is being 'capitalised' at the end of each year BUT the tick-size is NOT being increased accordingly.
This means that the performance attribution for a profitable strategy will decrease exponentially over time to eventually become meaningless.
By way of example consider the following:
ESignal default settings:
Initial Capital = $100,000
Number of Contracts = 100 (so I am effectively trading the strategy at $100 per tick of P&L)
Strategy Performance:
Year P&L
(Ticks)
2009 1,593.90
2010 1,277.12
2011 2,864.26
2012 1,005.59
ESignal Performance Calculations:
Year Tick Size Start P&L P&L End ROI
($) ($) (Ticks) ($) ($) %
2009 100.00 100,000 1,593.90 159,390 259,390 159.39
2010 100.00 259,390 1,277.12 127,712 387,102 49.24
2011 100.00 387,102 2,864.26 286,426 673,528 73.99
2012 100.00 673,528 1,005.59 100,559 774,087 14.93
You can see that whilst ESignal capitalises the denominator each year (i.e. we Start($) each year with the previous year's End($) amount) the tick size remains the same at $100 per tick.
I don't know if this is intentional or not but it is certainly not what happens in real life?
For example, if I am trading a fund which 'doubles' in the first year, then I will start trading 'double' the risk?
Accordingly, the performance profile using an adjusted tick size (adjustment = Tick Size * previous year's ROI) would thus look like this:
Year Tick Size Start P&L P&L End ROI
($) ($) (Ticks) ($) ($) %
2009 100.00 100,000 1,593.90 159,390 259,390 159.39
2010 259.39 259,390 1,277.12 331,272 590,662 127.71
2011 590.66 590,662 2,864.26 1,691,810 2,282,472 286.43
2012 2,282.47 2,282,472 1,005.59 2,295,231 4,577,703 100.56
In reality, the denominator is likely capitalised daily to accommodate subscriptions, redemptions, charges, fees and so on but this is obviously not necessary for the ESignal Back Test Report.
My suggestion would be to simply NOT capitalise the denominator at all, which will bring the monthly and annual performance figures back into line with what they should be.
In this respect, the performance profile using a constant tick size ($100.00) and a constant denominator ($100,000) -- the simplest solution by far for the Back Test Report -- would thus look like this:
Year Tick Size Start P&L P&L End ROI
($) ($) (Ticks) ($) ($) %
2009 100.00 100,000 1,593.90 159,390 259,390 159.39
2010 100.00 100,000 1,277.12 127,712 227,712 127.71
2011 100.00 100,000 2,864.26 286,426 386,426 286.43
2012 100.00 100,000 1,005.59 100,559 200,559 100.56
I hope this helps!
John
In the Back Test Report, I believe the "Monthly Returns & Drawdowns (%)" and the "Annual Returns & Drawdowns (%)" are being incorrectly calculated.
In short, the denominator used in the performance calculation is being 'capitalised' at the end of each year BUT the tick-size is NOT being increased accordingly.
This means that the performance attribution for a profitable strategy will decrease exponentially over time to eventually become meaningless.
By way of example consider the following:
ESignal default settings:
Initial Capital = $100,000
Number of Contracts = 100 (so I am effectively trading the strategy at $100 per tick of P&L)
Strategy Performance:
Year P&L
(Ticks)
2009 1,593.90
2010 1,277.12
2011 2,864.26
2012 1,005.59
ESignal Performance Calculations:
Year Tick Size Start P&L P&L End ROI
($) ($) (Ticks) ($) ($) %
2009 100.00 100,000 1,593.90 159,390 259,390 159.39
2010 100.00 259,390 1,277.12 127,712 387,102 49.24
2011 100.00 387,102 2,864.26 286,426 673,528 73.99
2012 100.00 673,528 1,005.59 100,559 774,087 14.93
You can see that whilst ESignal capitalises the denominator each year (i.e. we Start($) each year with the previous year's End($) amount) the tick size remains the same at $100 per tick.
I don't know if this is intentional or not but it is certainly not what happens in real life?
For example, if I am trading a fund which 'doubles' in the first year, then I will start trading 'double' the risk?
Accordingly, the performance profile using an adjusted tick size (adjustment = Tick Size * previous year's ROI) would thus look like this:
Year Tick Size Start P&L P&L End ROI
($) ($) (Ticks) ($) ($) %
2009 100.00 100,000 1,593.90 159,390 259,390 159.39
2010 259.39 259,390 1,277.12 331,272 590,662 127.71
2011 590.66 590,662 2,864.26 1,691,810 2,282,472 286.43
2012 2,282.47 2,282,472 1,005.59 2,295,231 4,577,703 100.56
In reality, the denominator is likely capitalised daily to accommodate subscriptions, redemptions, charges, fees and so on but this is obviously not necessary for the ESignal Back Test Report.
My suggestion would be to simply NOT capitalise the denominator at all, which will bring the monthly and annual performance figures back into line with what they should be.
In this respect, the performance profile using a constant tick size ($100.00) and a constant denominator ($100,000) -- the simplest solution by far for the Back Test Report -- would thus look like this:
Year Tick Size Start P&L P&L End ROI
($) ($) (Ticks) ($) ($) %
2009 100.00 100,000 1,593.90 159,390 259,390 159.39
2010 100.00 100,000 1,277.12 127,712 227,712 127.71
2011 100.00 100,000 2,864.26 286,426 386,426 286.43
2012 100.00 100,000 1,005.59 100,559 200,559 100.56
I hope this helps!
John