Pegged-To-Primary VS. Market Orders

Discussion in 'Automated Trading' started by walterjennings, Jun 30, 2009.

  1. Iv been doing a bit of research into figuring out if one order type has a bit of added value over the other. The tests are being done on a liquid equity (SPY) which normally has a 0.01$ spread. Both orders guaranteed a fill, trading 100 SPY at market is 0.50$ more expensive in ECN fees than trading a Pegged order (assuming adding liquidity). I don't have enough data yet to come to a conclusion, but was wondering if anyone on ET has experience researching or thoughts on this topic.


    Iv also noticed that some of my Pegged orders are removing liquidity, which is slightly confusing since it should peg to correct side of the spread.
  2. pegged orders are only useful if you can afford to wait. so, comparing it to an order type where you're taking liquidity won't tell you much about the order types in relation to each other (other than the obvious). BUT, it may give you some insight at the _strategy_ level, on whether you can afford to be patient.

    wrt to the rmv issue, this is common for pegged orders on tighter spreads and is very much ecn dependent. some use regional quotes to establish the bbo, and it's common to have lagging locked regionals on something like SPY resulting in a rmv vs add.
  3. I think they are comparable. I wouldn't say its a matter about whether I can be patient or not. Since I know both orders will be filled. What matters is the expected price on entry. What I am trying to figure out is the expected slippage on both order types given random entries on SPY throughout the day.

    Im trying to look at it just from price obtained +/- cost standpoint. So if you take a system which fires off both orders randomly throughout the day. Given a bid of X and an ask of X+0.01 at any random point in the day.

    Buying at MKT should result in an expected buy price of
    = X+0.01 +/- average slippage (dependant on market factors) + comm/ECN fee

    Buying at Pegged to Primary should result in an expected buy price of
    = X +/- average slippage (dependant on market factors, specifically volatility with pegged) + comm/ECN rebate

    Im curious to see which expected value is better if you were to take a random sample of 1000s of orders and send both at the same time. I'm up to 10 samples so far in my live testing. My hypothesis is that one will be better than the other (on average), ie not the same, from an expected entry price viewpoint wrt to X.
  4. i understand what you're trying to do, but i don't think you see my point:

    using random entries to benchmark two very distinct strategies (adding vs rmving) will only tell you how viable one is over the other... _when making random entries_. if you were doing the same type of random fill test comparing which ecn has better fill rates when either only removing or only adding, then i'd say you'd have a valid experiment. it would also be a valid experiment if you used your strategies _actual_ signals.

    what you're talking about though, is paying the spread vs not paying it, with zero context. there are a lot of factors that come in to play that are 100% strategy/portfolio dependent. eg. it would be better to mkt in on a fed report vs. pegging on the finishing leg of an index arb; likewise it would be better to be pegging when rtm'ng something with one way flow. so, given zero context with random entries, i would say it doesn't matter what route you choose because the results will be meaningless.

    also, you're NEVER guaranteed a fill with a pegged order. you DO NOT know that both orders will be filled. those assumptions alone should make you question the validity of this test.
  5. Agreed. the results of testing random entries would not apply to a different system with 100% certainty. I was just using it as an example of why I want to compare the two orders without going into the specifics of my system.

    The live testing I am doing is using my system. Though I do think the results of an experiment on random entries would be somewhat telling towards the relationship between the expected values of different systems. I have feeling that intrinsically on average Pegged orders have more 'value' than market orders (if you were looking at it from an information theory viewpoint), since one is using a provided service from the market makers. one is providing a service.

    Would you mind explaining in what cases Pegged orders are not guaranteed fills? This worries me, I assumed they sat on the book until the market ticks down through at least one level, then it would be filled. Which I think is a fairly safe assumption any time before 3:55 or maybe even later.