I finished a two week backtest

[Haptic]

Ok vets, this took me ages, my start date was June 30, 2016, and I ran it for 365 days. I didn’t run it mechanically, I played it out day by day, without looking forward to cheat. So it contains my mistakes as well as my lucky decisions I guess. So it is therefore non deterministic. My pairs were:

EUR/GBP, USD/JPY, EUR/JPY, EUR/USD, GBP/CHF, GBP/JPY, NZD/JPY, USD/CAD, GBP/USD, NZD/USD, USD/CHF.

Stats!

Time:
Days processed=365
Months processed=12.02
Trades:
Total trades=153
Profit trades=98
Loss trades=45
Profit trades cons.=14
Loss trades cons.=6
Trades / day=0.42
Trades / month=13
Profit trades / month=8
Loss trades / month=4
Max profit trade=34.00
Max loss trade=20.40
Income:
Net profit=318.45
Gross profit=634.94
Gross loss=316.49
Profit / month=26.49
Average profit=6.48
Average loss=7.03
Max drawdown=86.88
Profit factor=2.01
Return, %=31.84
Other statistics:
Max lot used=0.02
Restoration factor=3.67
Reliability factor=0.30
Profit probability, %=69
Loss probability, %=31

What do you guys think of this? I need to know!

@Rickster99 you have pulled me up before, sometimes citing conviction. Lets say I repeat this backtest, twice more, across two other years, and get similar results. If I then open a demo account, and after three months, it shows a similar equity curve, this should be enough to have conviction, should it not?

@tommor for the record, I have ditched every indicator off my charts apart from the double bollinger bands which I can’t part with for now. All the way through this test I was trying hard to read the price action, specifically to read the price action to come up with a reason for why I shouldn’t take the trade. I am starting to feel that for me at my skill level, that is how I might be able to make use of price action, as an exclusion criteria.

[QuadPip]

Wow! That’s an impressive undertaking. I will try this when I find some time.

Did you enter or exit trades even during your normal sleeping hours?

Did this experience improve your trading? What did you learn from it?

[Haptic]

I only entered or exited, or modified trades at the close of the new york daily candle

I’m still trying to sift through what I learned, so here it is in no particular order.

1. Its not uncommon for there to be no trades for 4-5 days then suddenly be presented with a few options at once

2. This strategy has a slightly inverted risk reward ratio, so the win % rate is extremely important! You really can’t afford to be wrong and take a punt on an entry that looks pretty good. It has to be just right or you should just pass on it.

3. My stops were previously at half daily ATR. I was just getting whipsawed constantly. Using wider stops set off the market, and not of ATR, has helped alot.

4. Get used to not caring about missed opportunities… I will have to practice this.

5. Once I’ve got a trade moving locked in profit, and trailing, there is no reason to worry about double exposure or correlations. So for example if I’m in profit and trailing long eur/jpy, and a good long usd/jpy trade comes along, thats fine.

6. I’m seriously now considering making a picture gallery of 200 images or trades I’ve taken, categorised into winners and losers. And trying to work my understanding of price action up off that, or at least study them for patterns.

7. Black/white Marabozus can be a really good sign of strength and can increase the likelihood of the entry and trade playing out in my favour. Conversely, long wicks, hammers, pins etc are a big red flag.

8. I don’t need ADX to read whether something is trending or ranging. It is quite frankly, obvious, the moment I stop lying to myself.

9. Trade invalidation can happen anywhere. For instance lets say after 24 hours, I’m up but only by 6 pips. That can be a sign that I should get out. After all, what I predicted would happen didn’t happen. I"m still flip flopping on this because I’ve seen it go both ways.

10. Ignoring resistance/support points is suicide when trading a strategy like this. It tends to dart in and out of the market, with the average trade being about 24 hours, and the long ones being maybe 3-4 days. But if I see price action approaching a SR point, I just stay away. And many times there is an opportunity cost to this as price trends right through the SR point. But just as often price ranges around there for days and stresses me out. I want to stay in the clear zone, and let other people fight over SR points.

11. This system is far more dicretionary, and far less mechanical that I hoped it would be. I just need to get used to that idea and learn to trade.

12. No exotics.

13. If I am already in profit on a major, and other trade setups come, it can be good to use a cross pair, to just lessen the correlation a little bit. I seem to be most comfortable with one trade open on the majors, and one trade open on the crosses.

14. I might need to remove the total return % for my next backtest, because I am addicted to staring at it, and I think it is hurting me.

15. I am not emotionally tough enough yet. I actually find backtesting stressful, and I am desperate not to lose on every click. Having said that, I actually enjoy backtesting way more than I thought I would.

[Rickster99]

I like what you’re doing. But you still have some points where you haven’t decided what to do eg point 9. Secondly are you trading technical trading or discretionary as it seems youre doing both. Is your drawdown really 86%? As your return is only 31%. You asked that if it should be enough to have conviction. But that is a question for yourself, can you sit through the drawdown, do you trust your discretion with real money, will you change your strategy halfway? All these questions need to be answered first.

[QuadPip]

Is your drawdown really 86%? As your return is only 31%. You asked that if it should be enough to have conviction. But that is a question for yourself, can you sit through the drawdown, do you trust your discretion with real money, will you change your strategy halfway? All these questions need to be answered first.

That drawdown is in pips not %, unless I’m reading it incorrectly.

[anon46773462]

Personally, @Haptic, I think that, although one could argue that one 365-day period backtest is not enough to prove anything, you have certainly studied this with enough discipline, consistency and thoroughness to justify now forward-testing this on a demo account. There is no reason why you couldn’t still run further backtests simultaneously, but I think you would benefit in other ways from now demo testing.

For example, you say:

I am not emotionally tough enough yet. I actually find backtesting stressful, and I am desperate not to lose on every click. Having said that, I actually enjoy backtesting way more than I thought I would.

In your backtesting you are not reading the price action in the same time duration as the real life. You are reading it like in a fast-forwarded video which is far more stressful than watching it play out, and taking decisions in real-time.

In addition, there are different psychological impacts present when literally waiting 24 hours for the next close. Price can do an awful lot in the meantime and it will be a supreme test to either switch off in the interim or watch-but-do-nothing!

Also, markets are not homogeneous in their movement characteristics and the future may well be different to the past in terms of, for example, volatility, extent of trends, etc.

I suspect that only carrying out further backtests (unless you are making changes to your method) might only serve to create more doubt and uncertainty about the reliability of your approach rather than reinforce it (especially when your decision-making includes a high degree of discretion rather than mechanical responses).

Running a demo will be in real time, risks nothing and allows you to experiment with other modifications that might come to mind as it progresses. And, when the time is right, you will have the confidence to shift seamlessly up to live trading.

But this is just my opinion!

[Falstaff]

Personally, @Haptic, I think that, although one could argue that one 365-day period backtest is not enough to prove anything, you have certainly studied this with enough discipline, consistency and thoroughness to justify now forward-testing this on a demo account.

100% agree

Ok vets, this took me ages, my start date was June 30, 2016, and I ran it for 365 days. I didn’t run it mechanically, I played it out day by day, without looking forward to cheat.

Actually - without knowing it - You ARE “sort of” cheating

The thing is that no YOU do not know - what the next few bars will bring - BUT the “machine” does !

Ergo - when you get bars close to the top of the page, your subconscious is aware that the movement will soon be “down” and vice versa. Whereas if you’re getting closish to the top of the page and the scale redraws - then subconsciously you “know” that an “Up” is on the cards !

How do I know ? - I have a mate who does it that way and indeed I used to myself. There’s no way of avoiding that bias that I know of and of course it is better than nothing - but it is quite a problem as far as I can see.

@anon46773462 raises a couple of other points which are very valid too of course - It’s amazing how well we can use Support and resistance on the larger time scales when you don’t have to sit and wait days / weeks etc for the “move” to happen !

Anyhow as the lad says - once you’re reasonably happy - fire on into the Future and see if it works on “demo”