What Is Daily Bias? (ICT)
Daily bias is the direction a trader expects the coming session to lean, decided before the open: up toward the liquidity above recent highs, or down toward the liquidity below recent lows. One common way to read it in ICT trading is where the previous day closed against the day before it.
A daily bias is a trader’s call, made before the bell, on which way the day is more likely to travel. The idea comes from the ICT method, where the question is less “will it go up?” than “which side’s resting orders will price reach for first?” This page explains how a bias is usually formed, turns the most common rule into something countable, and tests it on 33 years of SPY.
How it works
The bias answers one question: which side is today’s target? Above the market sit the prior highs, where buy-stop orders tend to rest. Below it sit the prior lows and their sell stops. The buy-side and sell-side liquidity page covers why those orders gather there. A bullish bias says the day should reach for the pool above; a bearish one says below.
It is built from the top down. Traders who use it typically start on the weekly or daily chart, read the direction of market structure, and ask where the nearest obvious pool of orders sits. Some add whether price is trading in the upper or lower half of its recent range, the premium and discount idea. The general method is on the top-down analysis page.
The quickest input is yesterday’s candle. If the previous session closed above the high of the one before it, buyers finished the day in control and the next target is taken to be that session’s own high. If it closed below the prior low, the target is its low. A close back inside the prior range is read as indecision, and many traders stand aside or wait for the open to show its hand.
Then the session tests it. In ICT terms, the early part of the day often runs the opposite way first, the manipulation leg of power of three, before the real move. So a bias is not an order to buy at the open. It is a filter that says which setups to look for once the session is under way.
The rule, stated exactly
Most descriptions of daily bias leave the rule loose, which makes it impossible to check. This page uses the version built from the prior day’s candle, because it needs no judgment and can be run on any daily chart.
- Bullish signal: the signal day’s close is above the high of the day before it.
- Bearish signal: the close is below that earlier day’s low.
- No signal: everything else. A close exactly equal to the earlier high or low counts as inside.
Three outcomes are measured on the following session. Did it trade above the signal day’s high? Did it trade below the signal day’s low? And did it close above or below the signal day’s close? A next-day close exactly equal to the signal close counts as neither up nor down, which happened 62 times.
The data is SPY, the S&P 500 ETF, in daily regular-session bars from 29 January 1993 to 25 September 2026. The prices are the traded prices, not adjusted for dividends. Every day that has both a day before it and a day after it is included, which gives 8,470 signal days.
A worked example
Start with a run of bullish closes in summer 2026. On Thursday 30 July SPY’s high was $742.45. On Friday 31 July it closed at $747.03, $4.58 above that high, so the bias for Monday was up and the target was Friday’s own high of $748.90.
Monday 3 August did what the bias said. SPY reached $758.58, clearing $748.90, and closed at $757.67, above Friday’s close. That close was above Friday’s high too, so Tuesday’s bias was up again. Tuesday 4 August reached $773.41, through Monday’s $758.58, and closed higher at $771.33.
Wednesday 5 August split the two outcomes. SPY traded to $776.85, $3.44 above Tuesday’s high, so the target was hit. It then closed at $769.79, $1.54 below Tuesday’s close. A trader who read the bias as “the high gets taken” was right; one who read it as “today closes up” was wrong on the same day.
The bearish side worked the same way in September. On Tuesday 8 September SPY closed at $765.96, $3.04 under the prior session’s $769.00 low. Wednesday took Tuesday’s low and closed lower at $762.40, again under the prior low, and Thursday did the same, closing at $757.83. Friday 11 September then reversed: its low of $763.60 never came near Thursday’s $756.64, and it closed $6.46 higher at $764.29.
These are hand-picked days to show the mechanics, not a sample. Six signals, five targets reached, four closes in the bias direction. The full count is below.
The original data
Across 8,470 SPY sessions, a close beyond the prior day’s range carried real information about the next day’s range. After the 2,668 days that closed above the prior high, the next session traded above that day’s high 67.9% of the time. Across all days the figure was 53.8%. After the 1,967 days that closed below the prior low, the next session took that day’s low 62.2% of the time, against 44.9% for all days.
| Signal day | Days | Next day took its high | Next day took its low | Next day closed up | Next day closed down |
|---|---|---|---|---|---|
| Closed above the prior high | 2,668 | 67.9% | 28.1% | 51.4% | 47.7% |
| Closed inside the prior range | 3,835 | 56.1% | 47.6% | 53.7% | 45.4% |
| Closed below the prior low | 1,967 | 30.4% | 62.2% | 56.2% | 43.4% |
| All days (the base rate) | 8,470 | 53.8% | 44.9% | 53.6% | 45.7% |
It told you almost nothing about the next day’s close. After a close above the prior high, the next day closed up 51.4% of the time (1,372 of 2,668). The base rate for any day closing up was 53.6%. After a close below the prior low, the next day closed down 43.4% of the time, under the 45.7% base rate.
Both gaps point away from the bias, not toward it. They are small, about 2 percentage points, and sit roughly 2.3 to 2.7 standard errors from the other days, close to the edge of what chance produces in samples this size.
They lean the same way in both halves of the sample, 1993 to 2009 and 2010 to 2026. The fair reading is no edge beyond the base rate for the close, and if anything a slight tilt the other way.
The plain close-to-close version shows the same thing. After an up day, the next day closed up 52.4% of the time; after a down day, it closed up 55.1% of the time. A day’s direction did not carry into the next close on SPY.
The group table is published as SPY daily bias summary, and every signal day with its next-day outcome as SPY daily bias days, so each count can be rechecked.
Demand for the topic is high. Of the 24,971 trading and investing videos in the corpus this site studies, counted once per video id, 16 have “daily bias” in the title, from 12 channels, at a median of roughly 103,000 views. That is 9.6 times the corpus median of 10,684.
Nine of the 16 passed 100,000 views. The top one, from TTrades, has 1,550,702, and 11 also carry “ICT” in the title. Three of the titles advertise a method with “no daily bias”, so not every teacher in this space treats one as necessary.
Why the high gets taken but the close does not follow
Part of the range result is geometry, not forecasting. A day that closes above the prior high usually closes near its own top. On those 2,668 days the median close sat 15.1% of the day’s range below the high; across all days the figure was 41.9%. When the close is that near the high, the next session needs only a small move up at any point to trade through it.
A small move up is not the same as a day that finishes up. Of the 1,811 next days that took the high after a bullish close, 555 (30.6%) still closed below the signal day’s close, as 5 August 2026 did. On the bearish side, 440 of the 1,223 next days that took the low (36.0%) closed higher anyway.
So the rule is best read as a statement about where the range may reach. That is consistent with how ICT traders say they use it, as a target for the day, but it means a bias alone does not support holding a position to the close.
The inside days sit in between: after a close inside the prior range, the next day took the high 56.1% of the time and the low 47.6%, both slightly above the base rate.
When it fails
The first failure is treating a range statement as a close statement. On this data the target was reached often, and the close still went the other way about a third of the time. A trade that needs the day to finish in the bias direction had no measured help from the signal.
The second is the reversal after a run. Signals cluster in streaks, as August and September 2026 showed, and the streak ends without warning. On 11 September 2026 the bearish bias was three days old, and the session never went near the target.
The third is order, which daily bars cannot show. A next day can take the signal day’s low first, stop out every buyer, then take the high late in the session. Both count as reached here. Knowing which came first needs intraday data, and this study does not measure it.
The fourth is the market and the chart. These counts are SPY regular-session bars. Many ICT traders use index futures or currencies, where the daily candle includes overnight trading and closes at a different hour, so the levels and the counts will differ.
The prices are not adjusted for dividends, so on the 136 next days that fell on an ex-dividend date, the payout pulls the traded close slightly lower. On the adjusted closes the rate after a bullish close moves from 51.4% to 51.9% and the base rate from 53.6% to 54.0%, which changes nothing above.
The fifth is the rule itself. This page tests one mechanical version. Traders who add higher-timeframe structure or a liquidity sweep filter are running a different rule, and it needs its own test before any claim is made for it, ideally one fixed in advance to avoid overfitting.
Related
The previous day high and low page counts how often SPY trades through and holds the two levels this rule is built from, within the same session. ICT explains the wider method daily bias belongs to and which of its ideas are genuinely new.
Power of three describes the session shape a bias is meant to anticipate, with a false move before the real one. And mean reversion is the opposite idea, which the close-to-close counts above lean toward slightly.
Treat a close above yesterday’s high as a clue about where today’s range may reach, not about where today will finish. Write the bias down before the open, write the price that cancels it next to it, and let the first hour decide which one you act on.
— Michael Whitman
This page is educational, not financial advice. Test every idea on your own charts before risking money.