[smabie]

What's really cool about Alpaca is the integration with zipline via pylivetrader. This means that you can take your backtested factor models and easily convert them to live trading. Also, they have pretty lax margin requirements. You can borrow 2x the cash you have, 4x for intraday if you're a pattern day trader, and the maintenance requirement is only 25%. And with zero commissions, suddenly it's possible for the average joe retail investor to actually run a legit factor strat. If you're only trading highly liquid securities, it could even make sense to run a intraday only strat at 4x leverage. Lets say you're trading stuff that has spreads around 4 bips, your lost returns are going to be around 10% per year from trading costs. That's certainly a lot and you'll need a kick-ass strategy, but that 4x leverage might let you get there.

[tosylate]

Do you have any reference on `legit factor stats`? I do not understand how leverage will be beneficial. Leverage will also multiply the potential downside risk after all? I would like to understand more how the trading cost would make up for it on average.

[pedrosorio]

> I do not understand how leverage will be beneficial.

Same here, if I understood the grandparent they were saying you lose 0.04% on every trade (from spread) and "in a year" that is 10% (250 trades?).

If you have a strategy that has expected returns of x bips per trade then you make an expected x-4 and with leverage 4 * (x - 4). Both have the same "sign", so if x-4 is expected positive, leverage just makes it higher risk-reward.

In fact with leverage you also have to pay interest on the loan, which seems to be at least ~18 bips (or 3/4 of that since the returns are on 4 * cash and interest is on 3 * cash) for a single day loan (far exceeding the "spread cost"): https://www.schwab.com/public/schwab/investing/accounts_prod...

It seems the only way leverage would help is if there were fixed costs to trading (which I assume there are but the GP does not mention).

[smabie]

Well you have 504 trades approx, but you're buying in and selling out, so you're paying the cost of spread once per day (252 trading days in a year).

Let's say you have have a strategy that has 7% return and 3% volatility. That's a pretty fucking amazing strategy. So you lever up to 4x and now you have 28% return and 12% vol. With your spread tax, you now have 18% return and 12% vol. Using their 2x leverage, you'll have a 14% and a 6% vol. So while your Sharpe ratio will be lower, your max return will be higher. And since you can't eat risk-adjusted returns, the tradeoff could very well be worth it.

[pedrosorio]

I was just reading their docs [1], and I got two things wrong above: - Schwab's overnight interest rate is actually ~1.8 bips. Alpaca's is even lower at 3.75% annual which translates into ~1 bip / day - If you liquidate stock the same day, you pay no interest

[1] https://alpaca.markets/docs/trading-on-alpaca/margin-and-sho...