Wow so much stuff going on I feel. I managed to get my backtesting software in sync by manually entering trades as they occured in real life. That way I can follow along and correct discrepancies as they pop up and enter trades manually where they were different. I did notice one difference. Two of my stock were sold yesterday and not so in the backtesting software. It turns out the trailing stop % does not really continue to be a percentage in the back testing software. It is determined to be value X as a % of the price on BUY and then remains the same through it's life time. The real life trailing stop does not work that way. I modified the code to be able to do it either way. I am now backtesting both approaches to see what I should adjust. I also screwed up a bit with yesterday's buys. I created a bracket order with GTC sell upper limit and DAY trailing stop. I thought I would put in a new GTC trailing stop on the next day. But now I cannot find out how to link the new trailing stop to the existing SELL LMT so that if one gets triggered the other one gets cancelled. That is rather annoying. Next time I will just create a bracket order with two GTC orders and adjust the % on the trailing stop on the second day. Also, my work is gearing up and about to get a whole lot busier probably from next week. I might not have so much time to spend on this from then on.
IB pointed me to a video about how to create a bracket order using an OCA group and that works, except when submitting the SELL LMT part of it I get a message my account has not been enabled for going SHORT. Need to wait what they say. In the meantime I have a SELL LMT and TRAIL STOP on some of the stock that are not one cancel all. I will have to cancel them manually I guess when the time comes or I will replace them if I do hear something from IB. As to % trailing stop or amount (AMT) I am going to use the AMT approach. I will determine the AMT as a percentage initially but then the AMT will stay constant through the entire time I own that stock. So I will enter it as an AMT instead of % on the broker site. Remember this is 2nd day adjustment only, first day I use a different (higher) %. In my backtesting I did not do a BUY LMT. It always bought at MKT. I implemented buy LMT now, but it does not perform as well as BUY at market. For real life trading I am going to have a fairly high LMT on my buys. Actually in backtesting even higher is better, but results start to taper off and I prefer a lower LMT. I said my work would get busier this week, and I still expect that. But this week will start late in the US with Monday being a holiday and I am in Asia so I do not expect anything until my Wednesday or after. That's about it for today. Still feeling a bit excited about neural nets.
No updates really.......... this is getting boring. I am like plus 20USD. Of course if I disregard transaction costs (50+ transactions) I am up by 70-80USD or more. On 4k that is not bad, 1.5-2% return in a couple of weeks. I am just thinking this is too slow and I need to increase to maybe 16-20K in market instead of 4K. I have a short vacation coming up, maybe after that.
Alright so a while ago I entered market at about 22K USD. I was not seeing much movement because any upward momentum was being negated by transaction costs, so I went in at about half the capital I am planning to invest which is 22K USD (out of 42K USD starting capital). The brexit caused some excitement. I was up 1000USD before Brexit and -600USD after Brexit. That is the current state of affairs. A bunch of shares got sold automatically through the stop on Brexit day. I setup some buys for today just now. This is a good test for my strategy and for my nerves too haha. I feel fortunate I haven't deviated from my strategy yet, however I did kind of guess a bunch of my shares would be sold off if Brexit actually happened. I could have sold them the day before if I followed my gut feeling, and prevented 5% loss or so. However, following my gut feeling is not the strategy I am following so I kept them and let them be automatically sold at their stops. Let's hope the strategy has redeeming value! So a good test in any case.
Alright I am still trading on this, but not much movement. I am up like 700USD or something since I started.
Hello Guys, That quantstrat strategy rode for a year or more at which point I decided I could not reconcile live behavior with backtesting. I did not lose or gain much, market did well during that time. Since then I have used other tools, Zipline locally and Quantopian online, tried Quantconnect as well. I found that in general the strategy "20-40% TLT and remainder SPY" works well from that. I could not reliably make any other strategies work well. I grew my investment account significantly by continuing to fund it and I was in TLT+SPY (well the Irish domiciled ETF alternatives) up to Covid-19. I also used and use SPY put options as insurance. I did well during covid-19, because of the treasury (TLT) and put options. During covid-19 I hand-traded a bunch with not much net effect. Portfolio grew with recent market upswing. Last year I also experimented with some machine learning, trying to paper trade futures like MES or oil. I did not see any alpha there. For now I am still (again) in SPY+TLT (the Irish domiciled alternative likes SPY5/IWDA etc..) with put options SPY. The TLT part is <10% for now, looking to buy back into it when prices go down. Anyway, I just wanted to provide an update here. Regards!