Backtesting Results
Here are the backtesting results I have so far:
# Trades: 37
# Wins: 28
# Losses: 9
W/L %: 75.7%
Avg. Trade Size (Pips): 17.6
Avg. Win Size (Pips): 18.1
Avg. Loss Size (Pips): 16.2
Pips Won: 507
Pips Lost: 146
Net P/L (Pips): 361
Maximum Losing Streak: 3
Maximum Winning Streak: 7
I think these results look pretty good, although they are a bit worse than the results I have obtained by paper trading the same strategy. I have some theories as to why this discrepancy might exist, having to do primarily with the fact that there is one discretionary element to this setup, namely that I can choose not to take the trade if it doesn't look right:
-- When backtesting, I tend to take a worst-case approach, meaning that I am less likely to pass on a trade. (In fact, I took almost every signal.) The idea is that, since this is trading past results, I don't want my knowledge of the price history to become part of the results.
-- In real time, I have prepared more thoroughly in terms of my identification of support and resistance zones. My preparation in backtesting tends to be much more cursory.
-- In real time, while I don't have access to volume information, I do have clues in the form of the "pace" of trading.
-- My paper trading results may just reflect a particularly favorable series of trades.
My guess is that the discrepancy between the paper trading results and the backtesting results are probably some combination of the above.
The obvious question in my mind at this point is whether the combination of backtesting and paper trading results I have obtained now justify increasing my position size from the current 10 cents per pip (which is mainly a marker used during the paper trading process to ensure that the entry and exit points I obtained were as realistic as possible). What I have in mind is to trade some semi-serious position size, like perhaps one dollar per pip, for another defined period of time, and then reevaluate the results.
I might also want to do some more backtesting, or replay testing of my setup before increasing my position size. I will be traveling this coming week, and won't be able to trade, so this may be a good time to mull things over.
# Trades: 37
# Wins: 28
# Losses: 9
W/L %: 75.7%
Avg. Trade Size (Pips): 17.6
Avg. Win Size (Pips): 18.1
Avg. Loss Size (Pips): 16.2
Pips Won: 507
Pips Lost: 146
Net P/L (Pips): 361
Maximum Losing Streak: 3
Maximum Winning Streak: 7
I think these results look pretty good, although they are a bit worse than the results I have obtained by paper trading the same strategy. I have some theories as to why this discrepancy might exist, having to do primarily with the fact that there is one discretionary element to this setup, namely that I can choose not to take the trade if it doesn't look right:
-- When backtesting, I tend to take a worst-case approach, meaning that I am less likely to pass on a trade. (In fact, I took almost every signal.) The idea is that, since this is trading past results, I don't want my knowledge of the price history to become part of the results.
-- In real time, I have prepared more thoroughly in terms of my identification of support and resistance zones. My preparation in backtesting tends to be much more cursory.
-- In real time, while I don't have access to volume information, I do have clues in the form of the "pace" of trading.
-- My paper trading results may just reflect a particularly favorable series of trades.
My guess is that the discrepancy between the paper trading results and the backtesting results are probably some combination of the above.
The obvious question in my mind at this point is whether the combination of backtesting and paper trading results I have obtained now justify increasing my position size from the current 10 cents per pip (which is mainly a marker used during the paper trading process to ensure that the entry and exit points I obtained were as realistic as possible). What I have in mind is to trade some semi-serious position size, like perhaps one dollar per pip, for another defined period of time, and then reevaluate the results.
I might also want to do some more backtesting, or replay testing of my setup before increasing my position size. I will be traveling this coming week, and won't be able to trade, so this may be a good time to mull things over.
Total Comments 9
Comments
|
Before increasing your size, I suggest you follow the program I suggested in The Journal and start out real small for real. Even if you're able to trade rationally, things do come up. The key is to analyze these unexpected events without coming apart.
A reminder: * Trade the plan using real money in real time, spending only what is absolutely necessary on "tools" (currently -- 2006 -- this is SierraCharts with an IB feed) and trading the minimum number of shares, contracts, etc., allowable. If your plan is not consistently profitable, go back however many steps are necessary to arrive at a potential solution. Recalculate your win rate and profit:loss ratio on a continuing basis. * If your plan is consistently profitable in practice, increase your size to what is a comfortable level, maintaining a continuous loop of re-appraisal and re-evaluation. When things come unglued, back up as far as necessary to regain your footing. |
 
|
|
|
DbPhoenix,
Either I'm not understanding your comment completely, or else I did not make clear in my previous Blog entry on Paper Trading that I was in fact trading in real time with a tiny trade amount (10 cents per pip) for the entire series covered by that posting. It is based on the encouraging results of my trading with very small amounts in real time, combined with the backtesting results, that have me wondering whether now is the time to gradually increase the trade size. |
|
|
|
|
Sorry, I obviously missed that. Then, yes, now's the time to increase your size. Will you be all-in, all-out?
|
|
|
|
|
Yes, this is an all-in / all-out setup, but as I mentioned elsewhere, I realize that it is not the ideal situation. I see this setup as a stepping stone.
|
|
|
|
|
Understood.
|
|
|
|
|
Been looking through the blogs of some other people, and came across yours. I was wondering, why isn't an "all-in, all-out" strategy the best method? I've read about pyramiding and scaling in and scaling out, but most of the time the net result was optimized by using an all-in, all-out strategy. Be interested to hear your views on this.
|
|
|
|
|
Looks very good,
 you're right that a larger sample size is necessary but a month or two should do. You're not dependent on your trading so no rush.One thing that you should keep in mind, if don't already do  , is that even if you make money day in day out, one undisciplined day can wipe out a lot profits. |
|
|
|
|
zeon,
I apologize for the late response to your comment, which I only read today. I'm still working for a living and sometimes my "day job" gets to be very demanding, and limits the amount of time I have for my blog. My answer to your question is simply that an all-in/all-out strategy forces an exit even when that exit is not justified by the price action. As an example, imagine you bought a retracement in an up trend, and your strategy forces you to exit even though price shows no sign of stalling, much less reversing. Yes, it is a "win", but it is not making the most of the situation. On the other hand, if your time for trading is limited, an all in/all out strategy may be an acceptable option if your setup appears to be profitable over time. To your other point, scaling in and out may be advantageous at times. Most of the arguments I've seen for and against scaling in and out assume that one either scales in, or scales out. But a trader is not limited to a stark choice between those two options. One can scale out under certain circumstances and scale in under others. That can potentially offer a compromise between the safety of scaling out, and the additional profitability of scaling in. It all depends on the strategy employed, but having more options available can be beneficial, in my opinion. |
|
|
|
|
sparrow,
Thanks for your kind comments. I certainly agree with all of what you say. So far, I have been quite disciplined about trading this strategy, partly because I know all too well where the other path leads. Spending time developing and testing a strategy has many benefits, one of which is that untested strategies seem much less appealing. |
|
|
Recent Blog Entries by FX_Cowboy
- Trading Results - Series II (06-15-2008)
- Trading Results - Series I (04-18-2008)
- A first look at actual trading results (04-13-2008)
- Backtesting Results (03-09-2008)
- Paper Trading Results (03-07-2008)
