I have read the following and I am new to this too, so you need to take this with a pinch of salt.
Large cap companies the market depth is not correct because large institutional trades don't show up in the list.
You need to take into account the prices that the buy and sell orders are at because some buy prices are way below the last price and some sells are way above. I consider them not serious buyers or sellers and discount them from my numbers. If my time frame is short I use 5 % above or below the price, if it is longer I use 10%. Any buyers or sellers outside this range I don't count in my analysis of the depth.
There are three ratios you can do with depth that I know of.
(1) The number of shares buy/sell
(2) The number of buyers/ sellers
(3) If you divide these two ratios into each other you can see a relationship to the average parcel size being bought or sold, this is an issue if you believe those trading large parcels know something or think they do.
I think the thing with market depth indicators is that it quite often happens before the price changes.
I also think the spread has some affect on this as a tool as well. I think the wider the spread the longer your time frame may have to be.
Yes you have to graph it and as the ratios change it means different likely outcomes depending on whether the price is rising or falling at the time.
I have a PDF on it if anyone is interested.
With small stocks there's definitely a correlation between total shares showing in the depth (B:S) and the SP. I'm sure there's edges available to exploit.
If you want to investigate your own edge, have a look at either Spark or *********** depth features.