My hypothesis regarding L2 analysis...

[Chagi]

Hello all,

I've been trading for about 6 months now, mostly longer term swing trading, but I've been making a point of spending plenty of time watching the L2 for specific stocks intraday. At first I found that when I was watching the L2, I had no real ability to predict where the stock would be moving. Another poster used the analogy of "Matrix code", and this is what I felt at first, in that it had no meaning for me.

Recently though I have found that monitoring the L2 is giving me a much clearer picture of what "might" happen. For example, there is usually a bias towards absorption of buy or sell orders by the MMs, which you can see by both the market depth and the bid/ask moves. Specifically I mean that after enough time spent watching the L2, you gain the ability to notice that orders of a certain type will quickly move the trading prices, while orders of the opposite type have greater difficulty moving the trading values.

For example, if the stock is likely to trend downwards intraday, you will usually see MMs absorbing very little of the "bid" side, so even relatively low selling pressure will tend to move the bid prices down more rapidly, whereas the "ask" side will apear to absorb "limitless" quantities of buy orders., requiring far more buying pressure (relatively) to move the ask prices upwards.

Any comments on this hypothesis?

 

[timcannell]

This sounds very reasonable - as a programmer I am thinking of coding some sort of level 2 indicator in an attempt to reconcile prints with disappearing bid levels - the rate at which bids disappear with selling at different price levels might be very useful info - while I suspect it would be impossible to match up all orders and prints it could be possible to identify sitauation when volume printed at a certain price is greater than the size at that level on level 2...

Hello all,

I've been trading for about 6 months now, mostly longer term swing trading, but I've been making a point of spending plenty of time watching the L2 for specific stocks intraday. At first I found that when I was watching the L2, I had no real ability to predict where the stock would be moving. Another poster used the analogy of "Matrix code", and this is what I felt at first, in that it had no meaning for me.

Recently though I have found that monitoring the L2 is giving me a much clearer picture of what "might" happen. For example, there is usually a bias towards absorption of buy or sell orders by the MMs, which you can see by both the market depth and the bid/ask moves. Specifically I mean that after enough time spent watching the L2, you gain the ability to notice that orders of a certain type will quickly move the trading prices, while orders of the opposite type have greater difficulty moving the trading values.

For example, if the stock is likely to trend downwards intraday, you will usually see MMs absorbing very little of the "bid" side, so even relatively low selling pressure will tend to move the bid prices down more rapidly, whereas the "ask" side will apear to absorb "limitless" quantities of buy orders., requiring far more buying pressure (relatively) to move the ask prices upwards.

Any comments on this hypothesis?

 

[Chagi]

This sounds very reasonable - as a programmer I am thinking of coding some sort of level 2 indicator in an attempt to reconcile prints with disappearing bid levels - the rate at which bids disappear with selling at different price levels might be very useful info - while I suspect it would be impossible to match up all orders and prints it could be possible to identify sitauation when volume printed at a certain price is greater than the size at that level on level 2...

Sounds interesting. Keeping in mind that I'm a business student, not a comp sci guy, and that I've only done a single programming course in University (Java), here are a few thoughts I have on why this might be difficult task.

One thing I neglected to mention in my previous post is that I generally find it to be easier to interpret L2 for lower volume stocks, as opposed to high volume. Because of this, if you were to experiment with this, I would probably suggest experimenting with something lower volume first.

The other difficulty I see is that this is quite subjective, and programming must be very specific. My suggestion would probably be to code something very small that doesn't even look specifically at the L2, but instead maybe code some form of "momentum" meter. Perhaps look at the number of shares traded at a given value (or range of values) before the bid or ask changes? I've been looking at alternative choices for direct access brokerages recently, and the Cybertrader Pro has some sort of momentum feature, though I think that theirs focuses more on events (such as a MM moving to the inside bid/ask).

Actually, momentum might not be the best word for it, since you can usually easily see momentum on a graph, but I would suspect that by crunching numbers there would usually be a bias towards shares traded more easily causing the bid or the ask to move.