I’m going to bed now. We will have much further to discuss on another version of the same concept with APAD, which only averages the total gross profit column over the total average drawdown of each trade.
I know you defined APD, as Average Profit, which in Wealth Lab, means Average Win to Loss, or Net Profit/the number of trades. I am going to show you that these are equivalent statements.
With APD, the Net Profit/Total Drawdown for Each Trade is identically equivalent to AvgProfit/AvgDD. That is, the APAD would only use the grossprofit/avgdd to get a better understanding of the potential upside.
AvgProfit := NetProfit / PositionCount;
AvgDD :=TD/PositionCount;
I saw this this way, that the way I calculated TD is as follows
//all this loop does is cycle through every single trade and add up for net profit all the gains or losses. TD is then summed up for each trade
for p := 0 to PositionCount - 1 do
begin
NetProfit := NetProfit + PositionProfit( p );
TD:=TD+abs(PositionShares§PositionMAEPCT§/100PositionEntryPrice§);
The TD part is just multiply the position size by it’s entryprice and number shares * it’s maximum adverse excursion percentage.
This is where I found the equivalency in what we were saying the APD was, and what I think is a much better measure of performance, that displays the magnitude of the gains in the calculation.
These are identical statements
NetProfit/TD as I’ve used them in this way.
Whereas, (NetProfit/positioncount)/(TD/positionaccount) are exactly equivalent, and it just takes basic algegra to see why NetProfit/td= (NetProfit/positioncount)/(TD/positionaccount)
We need to go one step further in this.
I propose measuring, not the net profit, but the Average Win’s Profit.
In this way, we will see how Gross Profit is comparable in magnitude to the Average Drawdown, so that we come to another measure of excess return that can be given by only using the average win’s profit, which I call APAD
That is, since, I hope, we are in agreement that the APD stat is numerically calculated in leyman’s terms as NetProfit/TD, don’t worry, that TD or total drawdown on each trade, then what we have said is that these are equivalent to (Netprofit/positioncount)/(TD/positioncount).
This is not nonsensical. The fundamental bias is in the use of netprofit.
The difference we need to make is in using only the Average Profit of each trade, not, the net total profit per trade of each system.
Thus, what we’d really get an idea of how much excess return a system can generate is equivalent to (Gross Profit/[the number of wins])/(TD/positioncount), where GrossProfit/[the number of wins] is the average winning trade over the (Total Drawdown of each trade). This will be exactly the same as GrossProfit, which are only calculated when the trade was a winner, over, TD, or GrossProfit/TD.
This new measure captures better the potential drawdown overcoming power of a system.
Since I already have a system that’s been confirmed on wl4.wealth-lab.com for the last 4 years as the most profitable system ever written, it is fair to say that a decent APAD, like a decent long run APD of 0.3 about is somewhere around 1, which is what Ross likes to see in his APD, but you’ll never get there long term if you’re only looking at the net.
So, for [LINKSYSTEM_35438029]
The APAD ratio for that same 5 year result comes to 0.95, which is a measure of how much the average profit is to the average drawdown of each trade. I think this a much needed improvement for the whole site.
Long + Short Long Only Short Only Buy & Hold
Starting Capital $100,000.00
Ending Capital $306,831.97
Net Profit $206,831.97
Net Profit % 206.83%
Annualized Gain % 25.19%
Exposure 6.73%
Number of Trades 6,562
Avg Profit/Loss $31.52
Avg Bars Held 1.01
Winning Trades 4,211
Winning % 64.17%
Gross Profit $369,798.31
Largest Winning Trades $757.77
Avg Profit $87.82
Avg Bars Held 1.01
Max Consecutive 171
Losing Trades 2,351
Losing % 35.83%
Gross Loss $-162,966.35
Largest Losing Trade $-738.52
Avg Loss $-69.32
Avg Bars Held 1.01
Max Consecutive 91
Max Drawdown $-17,560.03
Max Drawdown Date 10/27/2008
Max Drawdown % -5.27%
Max Drawdown % Date 10/27/2008
APD 0.3427 Calculated as NetProfit per trade/Total Drawdown per trade which is identical to the Net Profit / Total Drawdown of all trades
APAD 0.9549 Now, instead of NetProfit/Total Drawdown per trade, we have Gross Profit /Total Drawdown per trade or (GrossProfit/[number of winnning trades]/(Total Drawdown per trade that I used as TD/positioncount
Wealth-Lab Score 354.5361
RAR 374.2717
MAR 4.7762
Profit Factor 2.2692
Recovery Factor 11.7786
Sharpe Ratio 1.6241
Sortino Ratio 4.3339
Ulcer Index 0.8114
WL Error Term 4.8430
WL Reward Ratio 5.2004
Luck Coefficient 8.6290
Pessimistic Rate of Return 2.1891
Equity Drop Ratio 0.0000
K-Ratio 0.2402
Seykota Lake Ratio 0.0043
Expectancy 0.4895
Expectancy Score 626.9095
What does it mean? Since Ross is guessing that 0.4 is a decent number, I put that a long term system is probably good at 0.3, just based on this analysis and the history of the system.
The new number, APAD, measures how much Gross Profit a system can make to overcome it’s drawdowns. The difference we’ll discuss at a later time, but I think this is a good solution for us.
We already discussed this and concluded it made more sense to keep APD as it is, that new/similar measures didn’t necessarily help capture things any better, and decided instead the interpretation should be better defined as that seems to be where the disagreement lies.
I am not sure that I understand your suggestion. Am I correct that you propose to compute APAD as the average profit of winning trades divided by the average drawdown?
I agree that this has some advantages. A disadvantage is that a positive APAD does not imply that the system is profitable. The non-profitable trades might be big losses.
In a private e-mail to Jon I made a similar suggestion: To use only the x% most profitable trades in the numerator and the x% largest drawdowns in the denominator. That is, insignificant trades (trades with both small profit and small drawdown) are ignored. This has the same disadvantage as APAD though.
But perhaps this disadvantage is not so important. The only way to avoid it is to use the net profit, and there are already plenty statistics that use that as basis. Most people agree that one should anyhow use multiple statistics.
BTW, Hawk-fx has now an APD of 0.24. If you read the subscribers reviews this is the apex of hold n hope systems, and it once incurred a dd of 96%. Nevertheless, only 14% of systems have better APDs.
you need to boil this down. It is morning, and I do not understand precisely what this is trying to say, and I certainly cannot tweeze out how this is any better than current
In fact, I still have not seen anything yet that is better than as is. saying something is better does not make it so. Summing all max DDs and net profits was necessary since doing anything on a per trade basis is trying to assemble wildly tiny or wildly huge results that make the final results useless. But I don’t know if this was what you are trying to do here.
APD is sound and simple for subs to understand.
APAD as the average profit of winning trades divided by the average drawdown?
Yes.
The non-profitable trades might be big losses.
An APAD of 1 means your average profit is as big as your avg drawdown.
So a huge loss would be visible in the DD part then.
Still the other problem: if you have 99 trades with $1 dd and $1 loss, and one trade with $1 dd and $14 profit, then APAD = 14? But it lost $85 in total.
" APAD as the average profit of winning trades divided by the average drawdown? "
I do not understand why losing trades are ejected? The point is the summation of all trades, winners or losers, gives net profit. You are learning how much punishment (total of all max DDs) is encountered while en route to total profit (winners or losers).
???
We want to look at how large the Avg Profit is compared to the average drawdown to understand how well the system can overcome a drawdown, which I would value much more than the net effect of the system. Just as in WL where we have a “Recovery Factor” that is an index that gives a value on how quickly a system can overcome it’s drawdown, we need the APAD to tell us numerically how much of our average profitable trade can overcome the average drawdown.
I’ve surmised further that the APD can be broken down into
[ (GrossProfit-GrossLoss)/(NumberofTrades) ] / [ (TotalDDofeachtrade) / (NumberofTrades)]
In my case, APAD would just be
[ (GrossProfit)/(NumberofWinningTrades) ] / [ (TotalDDofeachtrade) / (NumberofTrades)]
Which is equivalent to saying GrossProfit/TotalDDofeachtrade
This way you see how large the win is compared to your average DD. An average win of 1 means your average win is about the same as the average dd. For a lot of systems that scalp, this absolutely will not be the case and you’ll see even more information by backing out the net effect of the gross loss. If DD continues to be large, this number will be very low, but bounded by 0.
I see it as an improvement because I feel APD with its net effects are long term statistics applied on a short term basis. The APAD can add more value to the community by explaining how large the average profit is compared to the DD.
SuperBands with an APAD at 1 means it’s very large on its average profit when compared to it’s average drawdown.
I know it would be a much needed factor. Rather than “hold&hope”, which might not actually be what the APD is capturing, the APAD will show how quickly a system can beat it’s drawdown. You said 0.4, I think, because you were guessing. From the superbands backtest of 0.3 and it being the most profitable script recorded on wl4.wealth-lab.com, we have a much better insight into what the actual decent APD ratio would be longer term. The APAD, I would say is good above 1, and, short of being a clairvoyant, you will never see a long term APD above 1, but you will with the APAD.
Here’s an example take from PTQQS
Realism Factor 89.2
Trades 55
Profitable 35
Losses 20
Win % 63.6%
APD Ratio 0.11
Avg Win $5,913
Avg Loss $8,204
To get Total Drawdown we have (355913-208204)/(TD)=0.11
0.11TD= (355913-208204)=389772.273=TD
My APAD, then is 35*5913/389772.273=0.531 est.
Let me get to a representative fund to get another statistic.