Last week we published the first part of EXANTE's Market depth module overview by Sergey Golubitskiy. Today we continue with it to get to the bottom of the topic.
The dark side of the Market depth
The previous part of this overview could have created a naive impression that the Market depth is an ideal instrument for market state prediction. It is that simple, however, only at first sight; if we dig deeper, its value becomes not that univocal.
Earlier we studied a prognostic model that has shown a perfect match. Is it enough, however, to say the model works, or is it a mere coincidence?
If we analyze some broader data, we will see that they are not that easily covered by the presented model. This is how our chart looks like (copied from the first part of the article):
First of all, the previous market turn took place at the level of 10,495, not of 10,466. The chart does not show any turns around 10,466.
Secondly, the weight of the orders is distributed in a rather strange way. In the first picture (first part) there were several hundred orders at the levels of 10,437, 10,442, 10,448, 10,462, 10,473. Their weight was almost the same as the weight of 10,445 and 10,466 orders, but they did not matter a lot.
Finally, the updates of the orders in the market depth is much more frequent than the chart fluctuations. The picture above shows that all orders at the level of 10,442 have already been executed, but the price has not dropped that low yet. On the contrary, there are a lot of orders over 10,447 that were made after the price trend turn (otherwise they would have been executed.)
Applying the logic we have used in the previous part, it is hard to explain why these 10,447-10,449 orders appeared. Sensible traders, according to our hypothesis, should have detected the horizontal channel and make orders at 10,445. They opted for upper price levels, however. Why? There are three major reasons.
The third view is worth being discussed in more detail. The orders placed on too low or too high levels can be placed whenever a trader wants, and they are not really likely to be ever executed. In such case such orders only indicate the wish of traders to move the price in this or that direction. Orders can be cancelled at any time before execution, so it is a good way to manipulate the market and move the prices towards the desired levels, especially when you use algorithmic trading.
This example showed this hypothesis works: while I was making the screenshot of EXANTE terminal, 1510 orders at the price of 10,447 disappeared (all at once), giving place to 2,000 orders at the price of 10,440.
Do not rely on the market depth too much
Now we know that the market depth, despite being a valuable source of information for the trader, should be used with caution, because only a part of the displayed orders reflects the real dynamics of the market. Moreover, it does not display a big portion of orders, Stop orders that are distributed against the Limit ones and can in fact influence the market a lot.
To sum up, market depth is a good predictive instrument, but only if you learn to read the information in the right way.