let's say your capital is allocated between strategies A and B. both strategies have exactly the same risk/return profile. on day 1, both strategy A and B enter 5 stocks each; assume day trade only. on day 2, strategy A triggers 10 stocks to enter and strategy B triggers 15. your capital is limited to a combined total of 15 stocks. if i limit the entry criteria, i.e. ensure that only a max of, say 5 stocks, are triggered each day per strategy, then there is way too much excess capacity, i.e. days when there are not many triggers. i am guessing it's a common problem with strategies where the distribution of triggers / events is clustered... - someone must have dealt with this - i'd appreciate any insights or references to relevant material.
Why not just trade the 15 stocks that you have the cap to trade? When you same limit the entry criteria did you mean change the entry criteria?
hi Imamic - when i test my strategies, i use a set of criteria / filters. the filters produce a different number of stocks to trade for each day. i trade systematically, which means that i take all the trades that my system generates - i can't just arbitrarily pick a smaller subset. what i am saying is if i limit my criteria to generate fewer signals, then too few signals will be generated, i.e. my capital will be idling too much. i think there is an analytic solution to this issue with allocation among multiple strategies. i just don't know what it is (yet).
Yeah you shouldn't let capital determine how you're analyzing possible setups. If you're satisfied with a strategy and you're getting too many setups, use another indicator or something to rank the setups. Like if you're buying/selling Moving Averages, when you get the long/short setups rank them by PE ratio. That way the pops you're playing, you picked for a reason. You might have a better chance. Of course it would be easy to test something like this.
great points and you are right - i have tested it this way, i.e. cherry picking the best trades on the busiest days - and guess what, it doesn't work; was a bit unexpected actually. the strategy works only if you take all trades as triggered. it's basically like this: you have a skewed coin toss, 55% to 45% instead of 50/50. there is 100 coin tosses on mondays, 500 on wednesdays, for example. you can't just cherry pick 100 of out 500 on wednesdays for it to work, you have to take all 500. the draconian choices are either to dump one of the strategies or to severely limit the criteria and then face excess capacity. i think there must be better solutions.
Okay. I'm familar with stuff like this. If you don't take all the trades, you have a higher chance of missing the winners etc. Yeah I've been in similar situations and I'm going to tell you what my boss told me: Move on. If it's not scalable it's not a good enough strategy. There are better ones out there and keep looking for them. Scalability helps out on commissions, slippage, execution errors, errors, and lack of capital.
ok thx but the strategies produce good returns, so there no reason to "move on". they are scalable, i.e. i can trade as much size as i want. my only question is how i optimize my capital allocation and hence return, among the strategies.
What's wrong with this picture? You come on here asking a question with an obvious answer about a sh1t strategy? Trade smaller. THat way you have enough capital to trade. And no it's not a good strategy if you can't "cherry pick" signals by filtering. Close Thread.
Reduce position size in each stock and take all the signals Position size = (Buying Power) / (Maximum number of positions expected in one day)
thanks for the input, but i am not sure how this is optimizing - for example: the sample is 10,000 trades, 10 trades per day avg. optimally, i want to have 10 trades a day and allocate 1/10th of capital to each. what you are saying is take max (say 100) and trade 1/100th. but then there will be days when i have only 10 trades triggered. based on my system, i will have to still trade 1/100th each, thus deploying only 10% of my capital, i.e. under-utilizing. systematically, varying the allocation is not "right", i.e. i would be altering the probabilities. hence, rather than varying the allocations within a strategy, i have several strategies and my idea was to figure out a more optimal combination of allocations between them.