FXperiment
Aspiring day-trader and currently broke college student explores learning the ins and outs of FOREX trading, both inter and intra day.
Justifile
Justifile
This is the concluding post to the 70/70 strategy re-analysis.
August 2011
win%: 50%
profit: -1.9%
difference: -2.3%
Here the anti-martingale is showing its weakness, with only two consecutive losses it failed to protect profits. When losses are not consecutive the winners get to be too small to make up for the losses.
September 2011
win%: 100%
profit: +7.56%
difference: +5.5%
Another textbook month for the 70/70 system.
October 2011
win%: 33%
profit: -3.78%
difference: -8.2%
The best month for the benchmark had two winning trades which worked unfavorably for the 70/70 strategy. October was the best month for the benchmark, so while this result is disappointing it is not totally unexpected.
November 2011
win%: 37.5%
profit: -1.02%
difference: +5.5%
Here is where the anti-martingale sizing strategy comes into its own. A series of successive losses with successively smaller lots limited the total drawdown to less than 4%, compared to almost 7% with the benchmark, with the total loss rebounding to only 1% thanks to the simple anti-martingale returning to regular risk after one winning trade.
End of Year Comparison:
Thanks to the compounding nature of the 70/70 strategy, which I couldn’t take advantage of with the benchmark due to simulator limitations (which I feel are not too far from broker limitations), and the anti-martingale position strategy, the 70/70 strategy has far and away exceeded the benchmark. In fact, with a total return of 24.3%, it came close to tripling the return over the benchmark. As expected, the overall win % is less, with the average (using monthly totals instead of total number of trades, which should be the same overall), being 63%, almost 10% less than the benchmark. However, this is still an edge over a market which, with proper money-management, can make significant returns.
By risking 2.1% of the account with each trade this strategy makes significant returns.
total year-end balance (with $10,000 start): $12,435.
Just as a note, I’m using a simple anti-martingale system for this analysis. For every loss I reduce position size by 1/2 of the previous, and after one win size returns to normal.
March 2011
win%: 60%
Profit: -.05%
Difference: -.25%
March is the first month which has performed poorer than the benchmark strategy. Although the benchmark was only making small profits, the extra loss incurred by the 70/70 strategy brought it down significantly. This is what happens with tighter stops, however, and the strategy is still performing significantly better than the benchmark.
April 2011
win%: 33%
Profit: -3.15%
Difference:-1.36%
Another month with an extra loss which brings 70/70 performance below benchmark performance across the same period. However, the anti-martingale strategy which has been applied to the 70/70 system helped a lot this month, saving more than 1% of profit earned in previous months. Where the benchmark was sitting at less than a 4% gain, 70/70 is better than 10%.
May 2011
win%: 33%
profit: -3.56%
difference: -.5%
Another poor month with a poor win percentage. However, it is performing about par compared to the benchmark. The benchmark had a losing month in May, as did 70/70, but while 70/70 only loses slightly more it gains much more. However, even the anti-martingale didn’t help here, as it was win loss win loss, which meant that the win was hampered while losses weren’t. These period happen and are unpredictable, but it is good to keep anti-martingale to protect against sustained draw-downs.
June 2011
win%: 100%
profit: +8.23%
difference: +6.2%
June exemplifies the strength of tight stops and defined profit targets. Where the benchmark allows profits to escape in the hopes of capturing a larger move and occasionally gets a larger than normal profit, the 70/70 strategy is able to make more money per-pip by keeping a tighter stop loss which wouldn’t necessarily work with the benchmark strategy because a 1/1 risk to reward ratio helps the 70/70 strategy, but tight stops reduce the win rate of the benchmark strategy. It may be worth testing a 70 pip stop on the benchmark, but it appears to be capturing fewer pips in its attempt to capture more.
July 2011
win%: 66%
profit: 1.26%
difference: -2.75%
A month with fewer trade opportunities, yet still profitable. However, July provided one of the benchmark’s few opportunities to make a larger than normal profit on a single trade. The benchmark made more than the difference between the two strategies on one trade alone. This is combined with a lesser win rate for the 70/70 strategy to give up money compared to the benchmark for July.
So this is going to be a quick and dirty rework using the same pip gains from the benchmark backtest which was completed last Monday. The goal is to see if a static stop-loss which is tighter to the initial price entered at, therefore allowing me to risk more dollars per pip, would make more money. I noted in my benchmark backtest that it was rare for trades to be 70 pips negative and then come back to a profit, so that is where the 70, in the “70/70” comes from. This strategy risks 70 pips in order to gain 70 pips. When a 70pip profit is reached the trade is closed. I’ll keep this to a month-by-month analysis. I will do my best to account for any possible compounding, which with such small lots is really only possible for every $1,000 in profit.
December 2010:
Win %: 80
profit: 6.3%
Difference from benchmark: +3.1%
Over five trades in December there were four winners. I assume each loser takes the full loss, which was not the case with the benchmark strategy. However, despite the full loss it was much more profitable than the benchmark strategy.
January 2011:
Win %: 100
profit: 6.3%
Difference from benchmark: 3.2%
There were only three trades, but each one was a winner. I’m already above $1,000 profit, so its time to start doing a little more math to deal with compounding. For every $1,000 I’ll add 10% to the profit or loss.
February 2011:
Win %: 66
profit: 1.2%
difference from benchmark: +2.9%
While February has proven to be poor for both the benchmark and the 70/70, the benchmark lost money and the 70/70 still made more than a 1% profit.
Three trades with one loss leaves a poorer win percentage, but one that is equivalent to the benchmark. Overall, the strategy showed a gain, which is good.
11.1.11
Sell signal. .05 lot. 400S/L
profit: 80.95
to date: 1551
@70: 1307.2
win%: 78.5%
11.9.11
Buy signal. .1 lot. 225S/L
profit: -224.50
to date: 1326.50
@70: 1082.7
win%: 76.7%
11.10.11 -> 11.13.11
Sell signal. .05 lot. 330S/L
profit: -122
to date: 1204.50
@70: 960.7
win%: 75%
11.13.11 -> 11.15.11
Buy signal. .05 lot. 300S/L
profit: -115
to date: 1089.50
@70: 845
win%: 73%
11.16.11 -> 11.20.11
Sell signal. .05 lot. 350S/L
profit: -60
to date: 1029.50
@70: 785
win%: 72%
11.21.11 -> 11.23.11
Buy signal. .2 lot. 120S/L
profit: -240
to date: 849.50
@70: 605
win%: 70%
11.25.11
Sell signal. .05 lot. 260S/L
profit: 10.75
to date: 860.75
@70: 675
win%: 71%
11.29.11
Buy signal. .1 lot. 150 S/L
profit: 21.60
to date: 881.85
@ 70: 745
win%: 71.4%
November turned out to be the antithesis of October almost exactly, with more than twice as many trades as October (indeed, twice the number of expected monthly trades), and a horrible win rate of 37.5% November gave up more than October’s total profit. This exemplifies the weakness of the system. There are times where the system is just too slow to indicate a move, as the moves have just finished when the signal appears. As has been shown across the last 12 months of data and a total of 49 trades, this is a limited period, but provided an enormous amount of the total pips returned to the market. Although the strategy is profitable as-is according to this test, I will be using the numbers generated to crunch a few possible improvements such as an anti-martingale sizing system, which could have possibly eliminated half or more of November’s losses.
10.7.11->10.10.11
Buy signal. .1 lot. 300 opening S/L.
profit: 191.80
to date: 1103.25
@70: 1062.20
win%: 77%
10.19.11->10.20.11
Buy signal. .2 lot. 120 opening S/L.
profit: 116.60
to date: 1219.85
@70: 1202.20
win%: 77.5%
10.26.11->10.27.11
Buy signal. .1 lot. 280 opening S/L.
profit: 250.20
to date: 1470.05
@70: 1272.20
win%: 78%
October 2011 exemplifies the qualities of this trading system. It caught significant moves with extremely limited screen time. This month provided a gain of more than 5.5% with a 100% win rate. The profit to date beat my goal amount with a month left by almost a full 5%. However, as good as October is, November exemplifies the weakness.
9.14.11
Buy signal. .15 lot, 140 opening S/L.
profit: 60.30
to date: 765.05
@70 852.20
win%: 75%
9.22.11
Sell signal, .1 lot, 250 opening S/L
profit: 90.40
to date: 855.45.85
@70: 922.20
win%: 75.6%
9.27.11
Buy Signal. .1 lot 170 opening S/L
profit: 56.80
to date: 912.25
@70 992.20
win%: 76.3%
September was another good month, but one short on trades. Another month without a losing trade, and a 2% gain put it right on target, although again most of the trades are small.
8.7.11
Buy signal. .1 lot, 225 opening S/L.
profit: 63.50
to date: 719.15
win% 80%
@70 918.50
8.12.11 -> 8.15.11
Sell signal, .1 lot, 220 opening S/L
profit: -112.80
to date: 616.35
win%: 77%
@70: 815.70
8.15.11->8.16.11
Buy Signal. .1 lot 200 opening S/L
profit: 111.50
to date: 717.70
win%: 78%
@70 885.70
8.22.11
Sell signal .15 lot 155 opening S/L
profit: -188.55
to date: 529.30
@70: 697.15
win%: 75.7%
8.28.11 -> 8.31.11
Buy signal. .15 lot 165 opening S/L
profit: -88.95
to date: 440.35
@70: 608.2
win%: 73.5%
8.31.11 -> 9.1.11
Sell signal. .2 lot, 120 S/L
profit: 264.40
to date: 704.75
@70: 748.20
win%: 74%
August was rough. With only a 50% win rate this was a profitable month, but only slightly. These happen, interestingly even an anti-martingale lot sizing strategy, rather than a static strategy as I have, would not have performed better. In fact, it may have done worse, depending upon the implementation. I will do an analysis after I finish the results post regarding the 70/70 strategy and an anti-martingale lot system.
7.6.11-> 7.7.11
Sell signal. .2 lot, 130 opening S/L.
profit: 330.60
to date: 587.1
win% 77%
@70 743.5
7.15.11 -> 7.21.11
Buy signal, .05 lot, 340 opening S/L
profit: 25.75
to date: 612.85
win%: 78%
@70: 778.50
7.29.11
Sell Signal. .1 lot 200 opening S/L
profit: 42.80
to date: 655.65
win%: 79%
@70 848.50
July was another good month. Another month without a losing trade, and a 4% gain make it rather spectacular, although again most of the trades are small fish, with one trade making up the vast majority of the profit. The gap between systems is closing.
6.9.11
Sell signal. .2 lot, 100 opening S/L.
profit: 138.40
to date: 193.7
win%: 74%
@70: 393.5
6.21.11
Buy signal, .1 lot, 300 opening S/L
profit: 32.20
to date: 225.90
win%: 75%
@70: 463.5
6.26.11->6.27.11
Sell Signal. .1 lot 270 opening S/L
profit: 6.40
to date: 232.3
win%: 76%
@70 533.5
6.29.11
Buy signal. .1 lot, 250 opening S/L.
Profit: 24.20
to date: 256.5
win%: 77%
@ 70: 603.50
June was a bounce back from May. Just less than a 2% gain is respectable, however, and the 100% win rate is nice. The trades have been for small gains however, and the @70 strategy is performing particularly well. Would have been a massive gain with a 1:1 Risk to reward ratio, allowing me to risk .3 lots per trade with a 70 pip target.
I’ll be making 6 posts soon for the final half of the year’s data. The result was an 8.8% profit, which is definitely something to work with, but I won’t go live until I reach 10% in a year. I’m close, but not there yet.
