Preliminary Results Show the Tick Size Pilot Study Affects Slippage

I looked at my slippage for almost all of my trades for this year with respect to the different groups in the Tick-Pilot Study.These were window trades. “Almost all” because I was lazy a few days.

As you all probably remember there are 4 groups counting the control group. The G1 group requires the prices of stocks be quoted in nickel increments but can actually trade at any value.

The rules for groups G2 and G3 are more likely to affect slippage, IMHO.

In the below attachments you will see the descriptive statistics for each group. The next two attachments are t-tests done with Excel.

G2 had two trades that were very significant outliers (excessively high slippage). “G2 with Outlier Removed” is just as it says: these 2 outliers were removed. I did a t-test comparing the controls and G1 in one group to G2 and G3 in the second group. I got significant highter slippage for G2 & G3 combined.

I worried about the outliers and compared the controls & G1 to “G2 Outlier Removed” and G3. The difference continued to be statistically significant. I show the results with and without the outliers removed

The results with the outliers removed showed a meaningful difference: 0.10% compared to 0.24% or a difference of 0.14%.The slippage is more than doubled!

Statistically speaking I conclude: WE ARE BEING SCREWED (if anyone is offended by that I will edit it: just comment). p = 0.02

I also added some trades from last year and the significance increased. But I was using a different universe last year and the mean slippage was affected. So I only reported this years results.

Weakness: The decision to combine the groups is reasonable but was not decided at the onset of the study.

Ideally, I would start a new study on Monday and report after a defined time-period such as the end of the year. I will combine the groups as I have done in this study.

If I do this I will likely be able the confirm: WE ARE BEING SCREWED!



Screenshot 2017-03-10 20.18.42.png

Screenshot 2017-03-10 20.19.01.png

Jim -

Since you’re using window trades, I don’t know how applicable this is to folks like me who use only limit orders. For one thing, the tick size program is supposed to improve liquidity, which means fewer partial fills, which means lower commission costs and fewer lost opportunities. I have no idea if liquidity has improved or not.

About half the stocks I buy are in the tick size program. But I haven’t made enough trades since its inception to evaluate whether it has raised or lowered my overall transaction costs. Fidelity, my broker, just lowered its commission by 37%, but it could be that my overall transaction costs are going to be higher because of the nickel price increments. The bid/ask spreads for the tick size stocks seem to be larger than they used to be.

  • Yuval

Here’s the preliminary results from a study in Trader’s Magazine.

Even in Tick Pilot Stocks, Liquidity Comes at a Price

Traders Magazine Online News, January 24, 2017

Colleen Ruane and Phil Pearson

The Tick Size Pilot Program has resulted in increased liquidity for included small-cap stocks—but that comes at a price.

ITG’s analysis, which draws on its proprietary Global Peer database of tick pilot trading by more than 100 institutional investors, finds that trading costs since the pilot launched in October 2016 are almost 50% higher, on average, compared with the control group. Costs for the control group are up 43% since the start of the tick pilot due to changing market conditions, but costs for Groups 1 through 3 in the pilot are up far more - an average of 91% since the start of the pilot (Chart 1).

With the Tick Pilot Test Groups accounting for about 4% of U.S. trading volume, this dramatic cost increase can put a drag on trading performance.

Here are some suggestions for managing the impact of the Tick Pilot on a trading P&L:

Favor algorithms with effective passive routing strategies; many brokers are struggling to execute passively.

Increase routing to inverted pricing (i.e., taker-maker) venues. One of the main changes we have seen in tick pilot stocks is a big increase in inverted venue usage, up almost 65% for the Test Groups (Figure 2). This is primarily because of large queues on the NBBO.

Avoid liquidity-seeking algorithms that don’t post and only cross the spread. With wider spreads and higher costs, these algorithms are more expensive, as there are not opportunities to trade with narrow spreads.

We recommend dark algorithms to take advantage of the increased dark liquidity in tick pilot stocks and spread savings at midpoint. Be sure your settings are set to peg midpoint.

While initial data from the tick size pilot suggest quite clearly that trading costs are higher across the board, overall the pilot has run relatively smoothly, with no major issues despite a large technology lift across the industry. Nonetheless, it is still in the early days of this experiment and we are going to continue evaluating the outcome to determine the pilot’s true costs and impact on market structure.

Now if anyone out there knows what the stuff about “passive routing strategies,” “inverted pricing venues,” and “dark algorithms” means, and whether that implies anything about how to actually trade these stocks (i.e. use market rather than limit orders?), please let me know.

Well stated. I like that you did not jump to any conclusion about the actual effect on liquidity.

I am not aware that the SEC is sharing any early results on their study. They will have a massive amount of data when they are done. Will they look at, and tell us, how much the average trader has contributed to this effort? Maybe. But today, I am thinking they could have bought me dinner before they… Well, sticking to the numbers.

In round figures that would not represent my situation so much. About 100 trades (in G2 and G3) this year at $10,000 dollars per trade with 0.14% increase in transaction cost. 100 * $10,000 * 0.0014 = $1,400. Annualizes to about $7,280 ($1,400 * 52/10 weeks).

No doubt they will send me a thank you card if they expand the number of stocks with nickel spreads at the end of the study.

I’m getting less inclined to speculate as I go along–just stick to the fact when possible. My data seems to say something about window trades. If I had to guess, one could probably extend my results to market trades. Or basically, any trade where a market-maker is pocketing some extra money with the new rules. It may be my physics background that makes me what to say there is conservation of matter, energy and probably money. But, also, conventional wisdom applies: there are no free lunches.

Don’t forget, the stated purpose of all of this is to get the market-maker a little extra money. Where is that money going to come from if it is not from the traders?

I’ll be interested in what you find when you get enough data—assuming you find this interesting enough to pursue.

One final word on this. Maybe I will notice, and be thankful for, the increased liquidity during the next “Flash Crash.” But for now, where would I look to find the benefit of this (for my type of trades)?

It is anecdotal—but I find it interesting—that the socks with the most extreme slippage (suggesting possible liquidity problems) were in group G2. These are the outliers that I removed.


I missed your second post during my edits.


I will have to think (and research) why G1 was affected in their study (compared to controls). Not what I would have guessed and not the case in my small study with my type of trades: but then, I know nothing about “inverted dark witchcraft” and how often it was used in their study population.


I just remembered something that Marco wrote when he first introduced variable slippage:

So for stocks in the tick size program, one would need to add (0.05 / Price) to the result. That can be huge. A low-priced stock like APT would get an extra 1.75% slippage added to every transaction, and because its average daily total is so low, transactions are already quite expensive.

(On the other hand, for stocks that trade below $1, you’d need to add only (0.0001 / Price) to the result. I’ve only just started trading in stocks this cheap, but it was very gratifying to get a limit buy order for TGD filled at 0.4104 when I couldn’t get it filled at 0.41.)

I wonder whether the variable slippage in our simulations should be altered for stocks in the tick size study.