Ah, that's more realtime fiddling than strategy fiddling. I'll do proper backtesting and forward testing of whatever I come up with, don't fret
Another 35 pips since 10pm :-D
Interesting. Having peaked at 400-pips just before 4 yesterday, it then crashed down to 70-pips or so (50 after commission). Now back up to 300. If anyone cares to look at the trades, and see if they can suggest a metric I can use to indicate I shouldn't be trading at say 4-6 yesterday I'd be very interested.
We've got two theories right now:
The fact that we get long strings of losing or winning trades suggests currencies overall go into trending/non-trending states, which intuitively suggests that's something I can detect algorithmically, but... eh, I think I may need more sleep!
We've got two theories right now:
- After the first losing trade, only paper-trade until the next profitable trade would have taken place. If you look over the trades, they're generally in long streams of losing trades. If it quits-early on the first loss and possibly block later trades, there's a good chance the damage can be minimised (rather than having a 350-pip drawdown).
- Use USD/CHF or EUR/GBP as a "canary"; both generally lose a lot of money with my strategy, so if they trade it's a bad indication about the market.
The fact that we get long strings of losing or winning trades suggests currencies overall go into trending/non-trending states, which intuitively suggests that's something I can detect algorithmically, but... eh, I think I may need more sleep!
Vrothdar
Well-known member
- Messages
- 344
- Likes
- 30
I think the idea of paper trading after a loss until the next win has merit. I've attached how your results would have looked it you'd applied that strategy. Best part of 280 pips bettter off.
If you're confident you do get your losses in runs then I think it's probably going to be your best strategy for cutting losses. You'd just have to be wary of short strings of losing trades cropping up.
If you're confident you do get your losses in runs then I think it's probably going to be your best strategy for cutting losses. You'd just have to be wary of short strings of losing trades cropping up.
The other issue is that it does simultaneous trades (trades are listed in the spreadsheet by the order they're placed), so it may well be in several losing trades before the first closes. That said, closer stop-losses and a few other tweaks and it could still trim a lot of that.
I'll see if I can get the spreadsheet output to produce these automatically later, but in the meantime here's a challenge for ya. On the attached sheet, have a look at what would happen if you remove all the USD/CHF trades, and all trades that open while a USD/CHF trade is open (where a trade starts within a few seconds of each other, assume the USD/CHF one was done first, as that's fairly easy to implement here). Doing the maths in my head, I make it a close to 200 pip profit-back. You lose a few profitable trades (hell, you lose some profitable USD/CHF trades), but also wipe out some massive losses.
Vrothdar
Well-known member
- Messages
- 344
- Likes
- 30
Stopping for lunch in a bit. Will have a look at it then. Need to put down the spreadsheet I'm developing for work first.
:edit:
By my reckoning it gains you 5 pips. Seems like cutting off losing streaks would be much more effective so long as missing out on that one winning trade here and there doesn't start causing problems if/when market conditions change.
:edit:
By my reckoning it gains you 5 pips. Seems like cutting off losing streaks would be much more effective so long as missing out on that one winning trade here and there doesn't start causing problems if/when market conditions change.
Last edited:
Yikes, yeah, I missed a bunch of them doing it by eye. Still, a small increase in profitability with a signficant decrease in market exposure is nice. I'll see if I can produce a sheet of what would have happened if it closed out all trades when the first loss hit, see how that comes out, but that won't be until tonight.
I've got my flatmate working on looking at other things such as time of day, momentum in USD/CHF when trading, average momentum, etc...
2 hours and I've barely scratched the surface of the analysis I wanted to do. Still, PDF attached that shows profitability over time... rather makes it tempting to try using MACD of profit to predict when I should get out fast! Still, better to never be in these trades, so probably will focus on entry conditions first.
Vrothdar
Well-known member
- Messages
- 344
- Likes
- 30
Still think your first idea of paper trading after a loss is more profitable and robust and far less mind blowing than your profit over time graphs! Hope you're getting more out of those than I am.
:edit: And no I haven't done any of the work I'd been hoping to do. Really struggling with my focus at the moment. Not just with regards to trading but in general.
:edit: And no I haven't done any of the work I'd been hoping to do. Really struggling with my focus at the moment. Not just with regards to trading but in general.
Last edited:
I get very little from the graphs, but at least I now know I get nothing from them. I'm taking a break before my eyes fall out, but I'll code up the spreadsheet to generate those paper-trade after a loss estimates next. In the meantime, that's what, 3 times the profit for half the trades or so? Quite promising indeed...
Here's the computer version. Much harder than I want to talk about to do, but done now. Turns out that actually there are so many trades in flight that the points in time it filters out are tiny. So far, 7 trades lost for a win of 120 pips. I think we have a winner...
Being nagged about implementing backtesting into my platform, so that'll probably have to be next-ish.
Edit: Nope, screwed that spreadsheet up too. Fixed, it drops about 20 trades, and saves only a handful of pips. Going to look at having it limit the number of positions it opens at once, and close trades out early if one loses.
Being nagged about implementing backtesting into my platform, so that'll probably have to be next-ish.
Edit: Nope, screwed that spreadsheet up too. Fixed, it drops about 20 trades, and saves only a handful of pips. Going to look at having it limit the number of positions it opens at once, and close trades out early if one loses.
Last edited:
Good to see someone progressing. Looks very promising for you good luck. Haven't had much time to move mine forward recently.
One idea for you rnicholl is to have your system monitor the trades and have some sort of shut off, so say you have 7 trades that all opened at about the same time and say 5 are currently losing more than half stop loss then on the balance of probability it would decide that a losing streak is underway and it's time to cut all trades.
One idea for you rnicholl is to have your system monitor the trades and have some sort of shut off, so say you have 7 trades that all opened at about the same time and say 5 are currently losing more than half stop loss then on the balance of probability it would decide that a losing streak is underway and it's time to cut all trades.
Rather than posting an Excel spreadsheet every few hours, quick update. It's now down to about 30 pips without the paper-trading after a loss, 130 with. That said, NinjaTrader estimates a 200 pip loss so far this week, so looks like statistics should save me in the end.
Good idea north5, we're looking at possibilites along that line. Not entering a third trade if there's already two losing in action being another. Or putting the stop-losses in tighter so that paper-trading triggers more easily. Or we might run a second version of the same strategy with less tuned parameters, and use its losses as an exit condition.
I tried paper-trading until two profitable trades, and basically the end result is just less profit. Good if you really want to get the trade count down, but otherwise...
Down to -22 pips now, with another 20 or so from commission. Having finally got the paper trading approximation system correct... err, it's at -95 pips. That said, it does a lot less trades to get there
Not impressed with Collective2; while there's some nasty losses here today, FXCM's prices (which Collective2 use) are INSANE if the market's moving fast. 80-ish pip spreads on major currencies. I can't see any way of using Collective2 for intraday Forex systems except if you exclude any time news might come out.
On a positive note, this saves me reworking for the trade exclusion stuff
Not impressed with Collective2; while there's some nasty losses here today, FXCM's prices (which Collective2 use) are INSANE if the market's moving fast. 80-ish pip spreads on major currencies. I can't see any way of using Collective2 for intraday Forex systems except if you exclude any time news might come out.
On a positive note, this saves me reworking for the trade exclusion stuff
