Dec

7

Slippage, from George Coyle

December 7, 2010 |

Does anyone have good thoughts or conventions on how to factor entry/exit slippage in futures markets? I am trying to find a general rule and google searches of academic papers result in not a lot and what I do find is a bit in excess for my needs.

My thought was to assign a rule such as when my entry/exit sizes combined total <0.50% of the average daily volume it would be safe to assume 1 tick worth of slippage per side. Ex. bid/ask 500/500.25 then assume sell/buy price at 499.75/500.50 (assuming a market that trades in 0.25 increments). Granted this would not happen in all instances but very likely not all fills would be at disadvantageous levels so the average slippage over time of this magnitude seemed fair. The scale could be increased as the percentage of average daily volume increased (i.e. <1.5% assume 2 ticks, <3% 3 ticks, etc.). A time of day factor might also be useful (i.e. pit vs electronic hours) as the liquidity is certainly better at peak trading times. Without very robust data this is difficult to test/determine so I wanted to see what the specs think.

Larry Williams comments:

It depends on the market and also on news of the day. Sometimes there's no slip at all other times it's massive so a decent trader would factor such things into his or her trading style; the problem is mechanical systems have trouble doing that which an individual can do.

Phil McDonnell writes:

There is no slippage with limit orders. Either you get your price or do not trade. If your testing can make a decision based on previous time periods. Then a simulated limit order can be placed. If the low for the next period is below the limit then you got filled on a buy. If the high is above the limit then you know you got filled. The tricky case comes in where the limit is exactly equal to the respective extrema. In that case I would suggest two ways of testing.
1. Assume 50:50 chance of a fill, take them at random.
2. Assume no fills at the exact bottom tick or top tick.

I recommend doing both tests. The 50:50 is probably realistic, but the no fills scenario tells you if the system might depend heavily on getting these top and bottom tick trades.

Larry Williams writes:

Phil's point is why I buy a small partial position on stop, then rest on a limit. This has helped my short term trading a great deal.

Jim Lackey comments: 

One should always be a gentleman here. How to trade is the easy part. How much to trade, when to trade and when to exit a trade are the difficult questions.

You can get partial fills, sub penny or fractional orders. Your limit is like a big round number for order sniffers HFT's other traders once made public. Everyone knows we should scale into positions.

When volatility is very low and liquidity is very high it's much easier to enter with out slippage.

The original mechanical question should be adjusted for volatility. I remember 2 times in year 2000 and again in the fall crash of 2008 where I used basically market orders with a price limit. Simply bidding through the current price to start + X ticks to start a scale or to end one. Back in those days the markets might rally 3% in an afternoon so you had to get some on.

Opens and closes are best to use at the market orders. However for the mechanics market at open or closes can be the best or the worst price of the day then your in either deep trouble all in on the high tick, a trend down or lucky you caught a trend day up.

The problem has been that since the markets have become more mechanical its the big up opens that continue to go up and the down opens that have gone straight down. These are order ladders and the moves are over by 10am. It was confusing to see the markets gap big and the entire day over by 10 or 11am. Now we are used to it…perhaps time for a change.

What is wild is to see the govie bond markets move a couple percent a day and the SNP move 1% and the ebbs and flows between stocks bonds and those indexing commodity etf's for investment over the next 100 years are guaranteed to lose.

Chris Cooper writes:

I often have the problem called "adverse selection" with my limit orders, in forex or futures. This is not a problem with small orders, but when your order is larger, you may (frequently) see a partial fill, then the market turns around in your favor and you make a profit. But the profit is not as large as it would be if your order had been filled to completion. The times when your order is completely filled, the market has moved through your price point and past it, usually. You are sitting on a loss if marked to market, and that loss is for your entire size. In other words, you get the bad fills when you really want to be filled, and the good fills when you don't want them (in retrospect). It's a big problem for short term trading.

Similarly, in markets with a long order queue such as ES, by the time your order is filled the market has pretty much moved past you and you are sitting on a loss. Unless you can play games with pre-positioning your order in the queue, you will always see this slippage with limit orders.

Jonathan Bower writes: 

You "could" avoid posting your limit order for bots, hfts, etal especially if it is of consequence (size) by working it semi-"silently" in an attempt to not get "screwed". One of the earliest execution algorithms was the "iceberg" where a limit order could be placed in (user defined) pieces, once taken out another one will replace it until your order is filled. Of course this is easy for those who want to to sniff it out. More recently you can add variance to your lot size firing in somewhat random orders as part of an iceberg. Or you could set a timed order to release a limit order (or combination of orders) based on a print, bid, or ask with what/if OCO ammendments, etc etc etc.

The bottom line if you need to be and can be creative with order execution especially in illiquid markets. Of course being right about the trade idea is still the hardest part…

Chris Cooper adds:

GLOBEX does not have liquidity that is as good as the forex interbank market, especially in the Asian hours. If you are trading smaller quantities, in New York hours, then futures are a better choice than forex.

If you are not getting filled when you should be, then your forex broker probably has what is called a "last look" provision in their system. This allows the liquidity providers a chance, say for 30 seconds, to decide not to fill your limit order. This has been an annoying problem for me in the past. The solution is to switch to a broker without this provision. It could also happen if your broker is not an ECN, and in that case you should find a new broker.

Jeff Watson adds:

There were (and still are) some people in the pit who made an excellent living with just the ability to tell when the market is going from quarter bid to quarter sellers. That, in itself is a huge edge.

 


Comments

Name

Email

Website

Speak your mind

1 Comment so far

  1. Ed on December 9, 2010 8:55 am

    Chris, if your strategy does not require you enter at a precise level, perhaps a market if touched order would work. You get slippage but at least you don’t load to boat only on the bad trades.

Archives

Resources & Links

Search