The easiest way to determine the effects of buying or selling on price is to look at an order book. The order book is the description of how much buying or selling it would take to move the price. The downside is that it describes the instantaneous changes in price, but not changes to the equilibrium price.
I'm a victim of being able to predict $10-30 moves, trouble is $10 moves are useless with bitcoin as you always get swamped by the fees...
I was thinking this too but it might be possible to use Google trends data or similar to work out if searches for the coin spike at that time. If they do, it's more likely to have fundamentals imo but I'm not entirely sure how you'd link the two anyway other than a higher volume moves a coin up on the list of coins on a lot of exchanges.
Researchers can use historical data (including Google trend) to a certain extent, but it should not be used for future prediction/projection/forecasting with 100% confidence. I mean, you will often find predictions differ from reality.In the mentioned paper, the researcher found correlations (not causation) between the absolute value of price change and volume. It means when the trading volume goes up, the price can either goes up or down. OP suggests that the x% volume -> x% price (causality), which is tempting, but I'd rather trust the good old 1987 paper.
Anacdotally for me, an external manipulation factor (either a person or a change of fundamentals) normally peaks the volume at the same time the price moves, this has often been a second long candle that pumps a coin really hard.
