This is an quiz question in linkedin's algo trading class.

Question: What approach would an algo trader use if the previously stable spread on ETF ABC widens against ETF XYZ from ABC becoming more valuable versus XYZ?

The answer is

Go Long ABC and Short XYZ

And the explanation is Yes, the algo trader would bet on the spread reconverging.

I have a hard time understanding the question. What's spread in this case? The bid-ask spread? If abc is becoming more valuable, which should I short abc if it's going to go back to normal?


I have an issue with the Algo answer but before I get to that, if the B/A spread widens, you have more to overcome just to break even on a position. In addition, you can't profitably pairs trade based on the B/A spread width because there's no meat there.

I have no idea what the lingo of the Algo class is so here's my take from having profitably traded an awful lot of pairs in every major volatile time period since 2008. You sell the overvalued security and you buy the undervalued security.

Something doesn't add up here (the question and their answer) so I'll leave it to you to figure out exactly what the Algo class is suggesting. Based on the wording of your question, I would buy XYZ because it's undervalued and short ABC because it's overvalued.

Here's an article about pairs.

| improve this answer | |

"spread" in this context is just the difference between the two ETFs (not B/A spread). Typically it's used to measure against some benchmark, but it can be used against two arbitrary instruments.

However, the answer seems backwards to the question. If one expects the spread between ABC and XYZ to close (meaning that ABC - XYZ gets smaller) then one would go SHORT ABC and LONG XYZ, not the other way around.

The answer does make sense if ABC trades for less than XYZ, in which case the "spread" is negative and you'd make the trades outlined in the answer (since ABC would gain relative to XYZ), but that detail is left out of the question as posted.

| improve this answer | |
  • What to short is dependent on over/undervalue not price (consider VOO at at $279 and SPY at $304). Whether the spread is negative or positive is also irrelevant. ABC-XYZ results in the same graph as XYZ-ABC only the direction of the change in value is different. It's the expansion or contraction of the spread that is the objective. Another consideration can be ratioing ABC versus XYZ to achieve a 1:1 dollar correlation. – Bob Baerker Jun 29 at 16:47
  • @BobBaerker Suppose VOO is consistently $25 below SPY (a "stable spread" as per the scenario), but at some point SPY rises to $310 while VOO only rises to $281 (the spread widens to $29). Which one if over/under valued? Maybe both are overvalued but the scenario is that an algo trader expecting the spread to retract back to normal would short SPY and buy VOO. So it is about price when you're talking spreads. I agree that in the real world the relative spread may be more constant but that's not the scenario that's lined out. – D Stanley Jun 29 at 17:19
  • Your example lacks the appropriate information to answer. Be that as it may, ABC and XYZ being over/undervalued as compared to underlying index is irrelevant. The pair is put on because one is overvalued as compared to the other, not to the underlying. A simple example with extreme numbers for ease of illustration. ABC is $50 and XYZ is $100. Both are SPX ETFs and they are properly valued. The next day, ABC is $55 and XYZ is $110 (both rose 10%). If you just looked at price (the spread widened $5), you'd be misled into believing that at $55/$110 it's a viable pair. It's not – Bob Baerker Jun 29 at 18:10
  • What's problematic in the above example is that if you put on the pair at $50/$100, a 10% move in the wrong direction loses money, yet the pct change was the same. That's why the positions need ratioing. As for under/over value, graphing either the difference in price or the quotient of price (Comparative Relative Strength) will depict the pair's over/undervalue relationship. It's more than I want to explain so I'll just state that if you tie the pair of the same underlying to yield, it clarifies the under/over value. – Bob Baerker Jun 29 at 18:19
  • @BobBaerker Again, I'm using the "previously stable spread" assumption in the question to mean a stable absolute spread, not relative, but the answer would be the same in either case. If the relative spread widens, you'd make the same trade. – D Stanley Jun 29 at 18:20

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.