I've been researching automated trading strategies on Index futures for some time now, but I feel I'm not making any headway.
I am restricted to trading broad Indexes/broad ETFs or FX due to the nature of my job.
I have a spread betting account with IG (alternatively I could use IB), which I use to record 1 minute bar index market data during the UK day though their API and record it for backtesting with later.
I'm a programmer by day and I have written a system for backtesting strategies in Python which gives me all the trades and statistics I need.
I need the system to be automated because I work all day and would like to start out slowly making only a couple of trades a day (or less).
I've tested all types of systems using indicators, all the way up to testing out some neural networks. I can't find anything that correlates well or is profitable. I feel like I've done quite a lot of reading and work but I'm not sure where to look next. I feel like that unless I have full level 2 book data and more I'm not going to find anything profitable.
Please could someone point me at a topic I should look in to when it comes to researching a mid-horizon strategy that might work on an index in 2016? I am not looking to make millions, I would be happy with one point a week at this stage!
B) Do you have a quantitative math background in addition to your programming background? If no formal background, how strong are your statistical and math skills?
C) Do you really need a model that heavily outperforms the index or do you simply need a yield management model that takes a base 7-10% return and adds 2-3% on top of it in which case a simple covered call strategy to a portfolio can get you that without too much stress and research. [Assuming you're save $10k per year and get 12% returns net of tax a year, you'll easily retire a decamillionare]
A simply issue you'll find looking at the index is the challenge of deciding what to do with the weight of the large over weighted stocks in an index. Take the US Nasdaq 100 as an example. If you take Google, Amazon, Microsoft, and Apple, those four stocks alone can single handedly cause that index to out perform either the Dow or S&P. And to extent by the same logic of cap weighting the DAX, CAC40, etc.
Now, if you feel those stocks will beat earnings forecast, do you given them more weight which results in more performance than the index or do you underweight trying to look for value if there is momentum being the stock causing it to be 'over valued'?
At what point do they become over/undervalued and how does that affect your own weighting in trying to decide the future price of the index?
I suspect you are being overly reliant on stock indicators which will not truly generate alpha that someone else has not already monetized before you. However, if you get your fundamentals right, then you can probably beat the index by itself by weighting components of your portfolio better than the index ETF.
Assuming you can answer the weighting portion of the question...then I'd move on to the next question after that in search of alpha.
A) I'm a programmer for a low latency prop trading firm, which means I know a lot about market data and trading at a high level. However what we do at work is quite far removed from what is possible for an individual trader. Also, I can't see any of the strategies, they keep it separate, which is why I'm doing these things myself from scratch.
B) I've not got a heavy mathematical background, when reading about topics I can generally follow along, although I don't always follow mathematical proofs. My statistical knowledge is not bad, although it's not degree level.
C) I certainly don't need to outperform an index, at the moment I'm just looking at any positive yield and want to work on building on that.
To answer your question about the stocks that might outperform their earnings. It depends on whether 'the market' thinks they will outperform their earnings, if it does then it will already be priced in to an extent in which case there is no upside in weighing those stocks heavily.
Doing a covered call strategy isn't something I've looked into before. However for trading individual stocks, or options on stocks, I require pre-approval and a minimum 60 day holding period. And if the stock happens to go on a restricted list then I'm unable to trade in or out of any position I have for an indefinite period. I can however trade indices and options on indices without any kind of approval.
Do you think it would work with the above restrictions?
Could you recommend some reading materials on fundamentals? I've not done much research on fundamentals and stock value for the above reasons, and to be honest can't really answer your questions with any kind of confidence.
(1) You should use the fact that you (presumably) have above-average programming skills and knowledge of market data to your advantage. For example, "L2 data" is relatively cheap as a retail subscriber - a hobby like tennis could cost just as much, and presumably the number of hours you've thrown at this problem are a lot more valuable to you than the cost of "L2 data". So there shouldn't be hesitation in using that. Aside from that, you can be creative with your sources of data; you could go after unstructured sources of data - flights/shipping data, Wikipedia API, 10-Ks...
(2) With the constraints you've listed, you can be a lot more creative with your options positions than your index positions. Instead of predicting prices or returns, consider predicting volatility.
(3) You do want to consider whether this makes sense for you. For a good portion of people who are already in finance and/or have their human capital highly uncorrelated with the broad market, the most sensible investment is just to go long the market rather than to seek alpha.
Assuming your goal would be simply to take the base equity index return and lop on a few extra points, it would seem to be that adding an options strategy on top should accommodation that goal and the instruments at your disposal (indexes and options on indexes is more than sufficient). As Artemiso has suggested, getting the volitility right would be a good way to go.
3. Yep I agree, this is what I'm trying to figure out. Ideally I'd like to beat my current investments in cash and a passive fund, hopefully it wouldn't be too hard to beat the 1% return I'm getting in a cash account.
2. You're right I should probably focus on some options strategies, I prefer the idea of predicting volatility as I'm not sure I want to rely on the index going up all the time.
Do you have any recommended reading regarding volatility prediction?
The reason I haven't bought any L2 data yet is because I'm not sure what I would do with a full order book any more than what I would do with any other data. If I had an idea I'd be there without a doubt.
Volatility is often mean-reverting. As a result, it is more predictable than compared with stocks, which do not exhibit a mean-reverting behavior. The Generalized Autoregressive Conditional Heteroskedasticity (GARCH [1,1]) model is a popular model for forecasting future volatility. You can implement the GARCH model in Python / R / Matlab, or even Excel.
Last edited by optionNinja; January 8th, 2017 at 04:01 PM.