Can you run both the 3% and the 1.5% in parallel, or at least let 3% run as is while you tweak the 1.5% model? The 3% model didn't run very long. Might be instructive to let it run till ruin or run for 200 sessions, whichever comes first. 3% may seem steep, but the timeframe was too short to know, coding notwithstanding, lol.
@TraderDaveUK， thank you for taking the time sharing this experiment with ET readers. While I am also working with a small team to program PA tradings, myself having both manual trading and industrial level software deployment experience, I appreciate reading your journal. Take my occassional suggestions with healthy doubts please. I suggest that you do something to measure the momentum of the market, and react to that too, and some of the reaction might be NOT to place a trade, yet. But the following suggestion might save you the trouble of developing algo to measure the momentum of the market, since momentum is a thing related to time/speed of the move and volume. Suggestion 2: since it is algo trading, you do not need to use minutes charts, which are still being used by human traders because manual trading need time to analyse, calculate risk/rewards, setup the trade and enter, thus minutes chart are necessary to give human traders the time to do all these. I suggest you switch to a tick or volume chart. 5 minutes chart gives you 81 bars a day (regular trading hours), but you can get around 10-15 times of bars if you are using a 1000 tic chart, depending on the volatility of the day. This way, you have more opportunities to trade, and also can skip more trades that are not optimal for your future tweaked algo. Of course, your stop loss and target can be smaller too, so that the 1000 dollars deposit will get you more testing, LOL.
Hello traders, @easymon1, thanks for your comments. To clarify, the risk was only 3% with the original balance of $1000. With an additional $1000 ($996.73 after commision) now deposited, the risk is reduced to arround 1.5% per trade on algo version 1.1. I do agree though, the algo has not run for any real length of time, and already I'm making adjustments - naughty. I do need to just let it run. @yc47ib, thank you for your constructive suggestions, I'm in the early stages of a scalping algo that uses lower timeframe data, so I will keep your points in mind. This past weekend I have worked further on algo version 1.2, and implemented an idea to increase the success rate, while handling false returns to the mean. So in 1.2, we'll be first filtering the opportunities to return to mean by price action (candle moving in our direction), taking those trades, but managing them by bundling them into a basket. That means in 1.2 we can have multiple trades open. Individually, they will have no stop loss nor take profit, but the basket will have an overall take profit and stop loss. The sample test balance/equity curve for 1.2 looks like this: You can clearly see where the basket stop loss is being hit, there's an average drop of 3.75% (min 2%, max 6%). However, a basket closes in profit nearly 80% of the time for an average gain of 1.5% (min 0.5%, max 5%). The downside of this approach, is that a basket may endure equity drawdown of up to 8% before the basket is closed at the average balance gain of 1.5% or a average balance loss of 3.75%. From today, I'm going to run both versions of the algo (1.1 and 1.2) - as a result you will see 2 charts in my end of day post, as a chart window in MT5 can only have one algo placed at a time, i.e. I'll have 2 chart windows, one running algo v1.1 and the other running algo v1.2, both in the same account. So, in summary, in case this has become a little confusing: 1. There will be 2 algos running, both apply the same principle of mean reversion. 2. v1.1 is a straight shooting, single trade at a time, Risk:Reward of 1:1.5, win rate 47%. 3. v1.2 is a more complex basket trader, taking multiple entries off of price action indications, average basket risk is 3.75%, average basket gain is 1.5%, win rate 79%. Let's see what today brings.
Hello traders, and welcome back to my world of pain. My trading day has ended, and the best laid plans, did not turn out to be the best today, take a look: Running PL: -$326.11 You see those four in a row losers from algo v1.1 to start the day - ouch, that was bad enough. But then, shortly after the first basket opened in algo 1.2, then closed for a -$95 loss, I realised that running 2 algos in one 'netting' account, is not going to work. Rather than being treated as individual trades, they are all thrown together. I am used to a hedging account, but admittedly naively here did not consider this significant difference and it's effect on this strategy of running both the algos. I thought I had it covered from the code perspective, each algo handling it's own trades, only closing it's own trades etc. But I see that the netting has basically negated that. So, deep breath. No-one to blame but myself here. Take it on the chin and move on. I've emailed AMP support to see if hedging accounts are offered for those outside of the USA. Regardless, to keep things working in this account, from tomorrow I'll be running the basket algo 1.2 alone. I'll keep the v1.1 algo on forward test via the strategy tester and update on that periodically.
I believe you can run on sim for a few weeks before you do serious damage to your balance. Just a suggestion.
You're down over 30 % in 6 calendar days, yes? Unless your account is play money which you don't care about anyway, I don't see why you'd want to have skin in the game right now while working out operational problems and errors in your approach which seems to be the phase you're in right now. Especially as you're trading algorithmically and emotions shouldn't factor in (one of the arguments against simulator trading). Either way - good luck.
Hi Laissez Faire, Thanks for your constructive and fair comments. The draw down is currently -15.8% due to the additional deposit, but I take your point. Yes, it's over $300 down in 6 days. Yes, I should probably have done some sim testing first. It's not quite 'play' money, but it is money I am willing to risk on the strategy I've put in place, and as I am now publicly committed, I'm going to see it through. I am confident the basket trader now running alone on the netting account will pull me back over the coming weeks. It is after all, an adventure. Not always plain sailing.
True. But it sucks to lose money from 'operational failure', learning a platform and fat finger mistakes rather than the strategy itself.
Again, fair comment. There is a pain threshold here. If we reach the point that I am facing a 30%-40% loss, then I reserve the right to shit my pants and re-evaluate.