Decimalization, liquidity, and why price-only charts miss the tape

On April 9, 2001, U.S. stock quoting moved from sixteenths of a dollar, or 6.25 cents, to one-cent increments. Decimalization is that shift, and it compressed the displayed spread from 6.25 cents to one cent. Spread-cost is the cash difference between the best displayed bid and offer, treated here as both a trading expense and an incentive to hide or drip size.

The same change reduced by a factor of 6.25 the economic incentive to rest size in the public book. When the spread was wide and time-precedence rewarded early arrival, resting orders were more fully visible, so large-player intent could be inferred from the book. Market-visibility is how much of true buying and selling interest can be inferred from the public book before large size is executed.

How large size left the displayed book

Once displayed depth thinned, accumulating or distributing large size became a process of slicing the order and adapting release to available liquidity rather than showing the full size at once. Algorithmic-trading is that automated slicing and timing, so large size can adapt to available liquidity instead of sitting fully displayed.

A liquidity-filter asks whether the book, spread, and available size make an intended order executable at an acceptable cost. Implementation-cost is the combined spread, impact, and adverse-selection cost of getting an order done, not just the last print.

Large-effective-volume as order-flow-analysis

Order-flow-analysis classifies short-horizon volume by whether price ticked up or down, then tracks those streams separately. A one-minute Barnes Group study over 80 days, using a rolling four-day window of about 1,500 minutes, classified volume as buying or selling only when price changed and discarded bars below a separation-volume cutoff. Separation-volume is the size threshold below which a bar is discarded when isolating larger participants from routine tape. Large-effective-volume is volume that both moves price from one short bar to the next and exceeds that cutoff.

In that Barnes Group window, large buying effective volume and large selling effective volume moved as near-mirrors, with waves of one side attracting adaptive waves from the other. The net difference between those large buying and selling streams stayed small. The archive reads that small supply-demand-imbalance as markets being generally well balanced even while both sides remain active.

What price-only reading missed

Applied to a continuous E-mini S&P 500 futures contract, the same large-effective-volume construction showed net selling from April 2008, well before a later decline, and that selling stopped near the start of July before an August reversal.

Volume-price-analysis reads price direction together with the volume that produced it, so a move is treated as a hypothesis about who is transacting. The archive argues that price-only technical tools lagged this microstructure change, that human discretionary execution cannot match automated analysis and routing, and that parallel-closed-markets can set a public print while large size transacts off the tape. Parallel-closed-markets are off-exchange venues where large blocks can print away from the public tape while the official price is still set in the displayed market.