This is a good question because it goes back to the basic principle in the back of the EV method and the LEV/SEV separation (Large/Small players).

The idea is that if a fund wants to buy for 5 Million$ of a $20 company that trades about 1 Million shares per day, the fund cannot obviously do that on the open market in one day. What sorts of possibilities exist?

There are the "dark pools" type of exchanges, which are fund limited.
The fund can also order these shares at an average VWAP price for the next X days. The order will go to the marketmaker who will try whatever possibility to keep the price low while accumulating shares for his customer (the simplest strategy is "stop fishing").
The next possibility is to buy shares on a regular base on the open market, but do it is small enough quantities so as not to push the price up. for example, the fund could buy 5% of the exchanged shares every day for the next 20 days or 10% for the next 10 days. That sort of volume will leave traces, because it pushes the Demand/Supply always in the same direction and the probability that you will see price upticks on large volume is higher than seeing price downticks on large volume. Therefore, by sorting the up/down ticks by volume size, if the majority of the large volume traded is responsible for upticks or downticks, then we can conclude that there was a "will" to buy/sell some large volume.

What we do not know is the reason for such a decision or the culprit. It might be 500 individuals who made exactly the same analysis and who bought each for $10,000 of shares (very unprobable) or more probably one ior two fund managers who bought a big chunk.

So, to come back to your question, the book's assumption is not that large players have left a trace. The assumption is that there is a trace of accumulation/distribution and when that trace is "strong enough" and occurs at critical period in the price trend (during a trading range, on a pull-back, in a support/resistance area), then most often than not, this gives the direction of the future price move. It is a statistical analysis, not a "sure pattern".


Pascal