Podcast: Franz and Ty talk TCA for FX
You can listen to the audio version here: http://bit.ly/13kzgL4
TD: Hi this is Ty Danco, I’m here with Franz Schmidpeter doing our latest installment of “Breakfast with BuysideFX.” In this case, I’m having my cup of coffee in Boston, while Franz is in our Burlington office. Franz, you’ve had an interesting week this past week I hear. You were doing some deep data diving with a money manager. How did that go?
FS: It went well. It was a midsize asset manager and he gave us a lot of data, all of his deals for one recent month and we dug very, very deep into it, spent a lot of time and had some very interesting findings.
TD: Now, the manager had some “lightweight” trade cost analysis before, comparing his FX executions to the daily range. I take it that that level of analysis really doesn’t uncover the real problems?
FS: No it doesn’t. TCA is one of those things where I always recommend you either do it right or you don’t do it at all. You shouldn’t do a half-baked TCA just for the sake of being able to say you have done it if it really doesn’t tell you anything. The daily ranges, even in the Europe session and Eurodollar right now, are an average of .75 and 1.5%, so that doesn’t really help you.
TD: Well, in this case, Franz, you knew the time stamping of the trade executions, and there generally are two approaches. One, you can look at the short view, like 10 seconds before and after trade analysis, or you can look at it in a broader context. What did the very short term show?
FS: We were looking at many thousands of trades, actually. And in terms of the larger ones, we were probably looking at the top five percent with volumes anywhere north of 20 million dollars equivalent.
TD: And so as in all things in life, it sounds like the old 80/20 rule applies. So, 20% of trades do 80% of volume. So, I take it most of the impact came from a minority of trades?
FS: Absolutely.
TD: So what did you find?
FS: Well, we found out that in the larger trades, there was a big discrepancy between the market at the Europe open and the market when the trades were executed. Now, you probably are thinking ‘well, there should be a 50/50 chance that the market moves up or moves down’, and that seemed to be the case for the small and middle trades. The surprising finding—or maybe not—is that in well above 90% of the large trades the market moved against the client.
TD: Hmmm. If you’re talking about thousands of trades being analyzed, I’m guessing that that’s statistically not insignificant. Are you giving us a conspiracy theory, Franz?
FS: No, it’s important that if you do a real TCA, you don’t have any conspiracy theories. We are just looking at a plain comparison of data.
TD: So if you have a big trade, you can move the markets? Do you think this was just because of this one manager’s large trades that the entire markets moved?
FS: No, in some of it lets say we had a currency pair like dollar Swissy and the exposure was pretty significant but probably not enough to move the market in and of itself. But there is obviously always the theory that some other people might have done something in the same country in terms of equities and fixed income on the very same day, and then people talk to each other and you might end up with a pretty significant multiplier effect, especially in the not-so-liquid currency pairs.
TD: So, you don’t have to come up with some theory of “oh, the custodian knows what’s going on, they’re front-running ahead of me.” You’re positing that when a market or security is attractive to one manager, another manager, especially one with the same style—especially if the manager has a quantitative style or momentum style, may be jumping on that same trade.
FS: Yeah. You don’t have to have an opinion, conspiracy or otherwise, on say the integrity of the Chinese Walls which should be in place at the custodians. We don’t want or need to start any conspiracy theories about trade practices, especially when we don’t have anything other than data.
TD: And we always know there’s “ALWAYS” a Chinese Wall built at that fax machine.
FS: Absolutely.
TD: So the manager did a good job and there’s not a whole lot to improve? What did you find out when you looked deeper?
FS: Well, to dig deeper for a real good TCA, we want to know what the market did not only between passing on the FX order, but what did the market do since the underlying equity on fixed income order. You have to look at market movement and risk as well when you talk about the cost of FX, just as we talked about this in our earlier podcasts on benchmarking. To do this right, you have to look at the timeframe between the origination of the FX exposure.
TD: So in this case, as is typical for most money managers, the FX was done at a minimum of several hours or generally the next day after the original securities trade which created the need. We’d think that traditional random walk or efficient market theory would tell us, that half the trades would be up and half the trades would be down and in the end, it all comes out the same.
FS: Yes, I think that’s the wishful thinking of a lot of people still, that sometimes you win and sometimes you lose, and FX is always a little bit like going to the casino and either putting your money on black or on red so it’s a 50/50 chance. What we found out—at least on this data set—that for the smaller and mid-size trades this 50/50 theory kind of works. Where it doesn’t work is on the larger trades. And even midsize clients have really some large trades to do and on all the larger trades, actually, to be precise, on 29 out of the 31 of the larger trades it wasn’t 50/50, but they lost, 29 to 2 That’s pretty remote.
TD: Hmmm. The odds of the dice coming up like that are something like 1 over 2^14th, 15th. That isn’t random. So what is the moral here? Why do you think this happened? 31 observations are not normally a lot of statistical proof, but my goodness, but it sure is significant when the normal odds are 1 in many thousands. Why did that happen?
FS: Well, let me reiterate we didn’t use any conspiracy theory or any speculative approach. We discovered this by taking the Europe open of the currency pair and comparing it with the rate where he executed. in 29/31 cases he would have been much better off if he’d dealt at the Europe open. Now, there could be a couple of reasons because, obviously, the custodian by itself has a Chinese Wall between the equity side and the FX side. But, there is always the issue that the underlying equity trade was not only done by this manager, but others like him. and market participants themselves often talk. So, you will always have a certain multiplier effect and that could run the market against the exposure.
TD: And that’s not going to be as big a factor in something like sterling dollar, but any exotic currency, this is going to be a bigger impact.
FS: Yes, you can be assured that the market participants who like to chat with each other will have a more intense chat in the less liquid currency pairs. I’m not even talking about exotics, but for example, something like dollar/South Africa, which is not a real exotic anymore, you can trade quite a decent size compared to what you could have done 10 years ago, but it’s still the multiplier effect in those currency pairs is much bigger than it is in Eurodollar or cable.
TD: So how big was this “time effect” compared to the “less than the best execution” effect? Or in other words, how much did time, which is manageable, or size—which can also be managed by different techniques, impact vs. simply poor or non-competitive, wide bid-ask executions?
FS: Once again, before I really mention the number, because it’s kind of a scary number, we only compared the rate at the Europe open—for this European manager—with the execution rate and we always took the same time. We never did anything like say “oh, this looks worse if we take it an hour earlier or an hour later, we always took the same time on those 31 trades. And the cost of the delayed execution was three times bigger than what you call the “bad execution” short-term cost, even when adjusted for size.
TD: And this is measured only on the next day, that is, the FX done generally at the start of that same day, which is actually T+1 from most of the original days, and doesn’t even include any other “getting ahead of the curve” benefits that could have been done if the FX had been done for T+3 at the same time as the originating equity trade?
FS: yes.
TD: Okay, so it sounds like what you’re saying Franz is not just from the risk-adjusted basis of the market going down, but of actual performance, it pays to get done early.
FS: Yes, because you always have to be assured of real, legitimate markets. Once you do an equity trade or a fixed income trade, you can never be assured that no one in FX market knows about it and will try to kind of front-run your exposure.
TD: So, once again, we find that speeding up the “pre-trade” preparation is probably the most important performance enhancer you can have, for the quality of the trader’s life, but also to the quality of client’s trades. You can’t talk about improving buyside trade costs until you talk about improving pre-trade processing.
FS: That’s right. You know, as traders, we can make judgments on whether to hold up or not based on our market knowledge or feel, but you know that the longer you wait, the worse the odds are that you win.
TD: And on that note, time to sign off until next time for Breakfast with BuysideFX. Thanks for listening
Tweet