To be clear, I was only claiming in my above post that my buy orders should be higher urgency at the open. I looked at this strategy’s orders, and its average closing rank is 96 – on average, stocks at that rank return 0.14% from Monday’s open to close over the in-sample, so I would be better off trading those with less urgency or towards close if I expect these results to hold up. It’s a high turnover strategy, about 14x/year, so that change would have yielded up to 2 percentage points of excess return over the in-sample assuming similar transaction costs.
I trade a different, lower turnover strategy in my IRA, but there I have to execute sell orders, at least partially, before I submit buy orders to open new positions. Presumably, some of your trading is in a tax-sheltered account. If so, how are you sequencing those trades? Assuming your account is fully invested, some of your buy orders would have to be either: a) delayed to a subsequent day, b) submitted later in the day as sell orders liquidate, or c) submitted for smaller amounts and modified up as sell orders liquidate.
It doesn’t surprise me that your transaction costs for VWAP orders are lower than your “non-VWAP” manual order working strategy, but it’s possible your transaction costs could be even lower with other non-VWAP algos.
That’s not a knock against VWAP orders at all. I think it’s a fantastic default for many cases, and I think most users would be better of using them over manually working orders, in both time and lower costs. Brokers employ teams of researchers and developers to build these execution algos and smart order routers to lower transaction costs and market impact, so why not avail that as much as possible. That applies doubly so if you’re trading through GS.
That said, in addition to exploring Fox River algos through IBKR, I’m also going to experiment with the VWAP and TVOL algos in my Fidelity account. Looking at the most recently quarterly SEC 606 filings from IBKR and Fidelity, it looks like Fidelity does a better job at sourcing off-exchange liquidity. For 2023 Q4, IBKR routed ~80% of non-S&P500 orders to exchanges while Fidelity only routed ~40% of non-S&P500 orders to exchanges. That’s a stunning difference, a lot of which is due to the PFOF providers like Citadel that Fidelity has agreements with, but Fidelity also executed 12.8% of those orders in their ATS while IBKR only executed 3.7% of orders in their ATS. This seems consistent with my anecdotal experience of the IBKR smart order router doing a poor job at sourcing liquidity for lower liquidity stocks.
Good question! Almost all my trades are in tax-sheltered accounts with no margin. I keep about 2% of my accounts in cash so I can avoid having to sell everything first and then buy later. But sometimes that’s not enough, so I do place my sell orders before my buy orders. Often I’ll place a buy order late in the day once I have enough cash to get it filled.
Yes, Fidelity does a great job sourcing off-exchange liquidity. No matter what algo I use, my fills at Fidelity are consistently better than those at IB. I’ll have to see what they’re like at Goldman . . .
My problem is most of my stocks have a 100% margin requirement, effectively removing the benefit of using margin. I suspect many of you have the same issue although I’ve never seen it discussed. Even though each broker may set their own policy for what they consider risky, I wonder if there is a general rule that can be implemented on the P123 side to filter out such stocks.
Different brokers have different margin requirements. IB, for example, has 100% margin requirement on almost all stocks with a market cap below $400M, almost all Canadian stocks, and almost all European stocks. Fidelity, on the other hand, has 100% margin requirement only on truly tiny stocks. I’ve looked now at a lot of different brokers, including Pershing, Wedbush, and Goldman, and none of them have margin requirements anywhere near as high as IB’s.
Some good news on this front: IBKR now supports execution algos for Polish stocks. I went to close out one of my Warsaw-listed positions today and saw the typical ex-US set of execution algos available from IBKR Mobile. I sold the position using the Adaptive algo with no problems.