I almost scrolled straight past it. Three tiny crosses on the GBPUSD daily chart at the top of a long climb, each one opening and closing at almost exactly the same price, the middle one floating a little above the other two like it had lost contact with the market altogether. My first instinct was to call it chop and move on. Chop, after all, is what three small candles usually are. But these three were not ordinary chop. They sat at the end of a 400-pip advance, they had gaps between them, and the market had just spent three whole sessions failing to make any progress at all. The textbooks call that shape a tri-star, and they call it a reversal. On a funded account, a textbook label is not a reason to trade. A pattern that is this rare, and whose published hit rates sit barely above a coin toss, has to earn its place with strict rules, a confirmation close, and a position size that keeps one loss small. That is what this guide does.
In this article
- What a tri-star actually is
- The three dojis, one by one
- Why three dojis can mean a trend is finished
- My funded-account rules for trading it
- The sizing maths under a daily loss limit
- A worked trade, candle by candle
- What twelve strict tri-stars actually did
- How it differs from the patterns it looks like
- The mistakes that turn a rare pattern into a loss
- My daily routine for a pattern this rare
- How it fits funded rules: drawdown, consistency and patience
- Frequently asked questions
It lays out the three dojis exactly, the confirmation close I refuse to trade without, the sizing maths on a 100,000-dollar account with a 5,000-dollar daily loss limit, a worked short trade with real prices, and what twelve strict tri-stars actually did when I checked them against the rules in this guide. It also says the uncomfortable part plainly: in a 24-hour forex market, true gaps are rare, so strict tri-stars are rarer still. Trade this pattern as an exhaustion warning that must be confirmed, never as a signal on its own.
What a tri-star actually is
A tri-star is a three-candle reversal pattern made entirely of doji candles. A doji is a candle whose open and close finish at almost exactly the same price, so the candle has almost no body at all, just wicks above and below a flat line. One doji after a trend is a pause. Three in a row, with the middle one gapped away from its neighbours, is a market that has stopped believing in its own trend three sessions running. The classic definition, from Steve Nison’s candlestick work and carried into the TA-Lib pattern library, demands three consecutive doji in which the middle doji gaps in the direction of the trend, away from the first and third. That gap is the whole point. Without it, you do not have a tri-star. You have three small candles in a sideways market, which is a different thing entirely and means nothing.
There are two versions. A bearish tri-star forms at the top of an uptrend: the middle doji gaps above the first and third, and the reversal, if it comes, goes down. A bullish tri-star forms at the bottom of a downtrend: the middle doji gaps below the first and third, and the reversal, if it comes, goes up. Both versions need a real trend before them. A tri-star in the middle of a range is not a reversal pattern, because there is nothing to reverse. I require at least ten candles of clear trend, with the move worth at least 150 pips on a daily chart, before I even look at the three dojis.
| Version | Forms after | Middle doji gaps | Reversal direction | Confirmation close |
|---|---|---|---|---|
| Bearish tri-star | Uptrend, 10+ candles | Above the first and third doji | Down | Below the third doji low |
| Bullish tri-star | Downtrend, 10+ candles | Below the first and third doji | Up | Above the third doji high |
The three dojis, one by one
Each candle in the pattern has a job, and if any one of them fails its job, the pattern fails with it. I check them in order, and I am strict about the measurements. A body is small only if it is tiny compared with the candles around it, not merely small in absolute pips. My rule is that each doji body must be no more than 20 percent of its own candle range, and no more than a quarter of the average body of the previous ten candles. Anything bigger is a spinning top, not a doji, and spinning tops do not make a tri-star.
The first doji is the warning. After a long trend, the market opens, wanders, and closes where it started. Buyers and sellers have fought to a draw for a full session. On its own that means little. Trends pause all the time. The second doji is the strange one. The market gaps in the direction of the trend, opens away from the first doji, and then closes where it opened again, stranded at a new extreme that nobody defends. In a bearish tri-star this middle candle sits above the other two. It is the candle that tried to extend the trend and found nobody willing to follow. The third doji gaps back against the trend and closes flat again. The market has now had three sessions to continue the trend and has failed three times, and the third failure comes with a gap back towards where the trend started. That sequence, trend, stall, stranded extreme, retreat, is what makes the pattern worth watching.
| Candle | What it must do | What disqualifies it |
|---|---|---|
| First doji | Open and close almost equal after a clear trend | Body wider than 20 percent of its range |
| Middle doji | Gap in the trend direction, then close flat at the extreme | No gap, or a body that overlaps the first doji |
| Third doji | Gap back against the trend, then close flat again | A strong close in the trend direction |
| Confirmation candle | Close beyond the third doji in the reversal direction | A close back inside the pattern range |
Why three dojis can mean a trend is finished
The logic is exhaustion stacked on exhaustion. A trend runs on conviction: each session, one side wins and price closes in its favour. A doji breaks that run, because neither side wins. Three dojis in a row break it three times, and the gaps between them show that even the market’s attempts to restart the trend cannot hold. The middle doji is the key evidence. The market paid up to a new extreme, gapped there, and then could not build on it for a full session. When the third doji gaps back, the trend’s last supporters have quietly left. What remains is a market with no conviction in either direction, sitting at an extreme, waiting for a push. That push is the confirmation candle, and without it the pattern is only a warning.
There is a second reason this matters on a funded account, and it is the honest one. Published testing of this pattern is not flattering. Thomas Bulkowski’s encyclopedia work puts the bullish tri-star at about a 60 percent reversal rate and the bearish version at about 52 percent, barely better than chance, and both versions rank among the rarest patterns in his catalogue. That is exactly why I never trade the pattern alone. Its value is as a filter at a level I already care about: a tri-star at a daily resistance level, after an extended climb, with a confirmation close, is a different proposition from a tri-star floating in the middle of nowhere. The pattern tells me where exhaustion is. The level, the confirmation and the size tell me whether it is tradable.
My funded-account rules for trading it
These are the rules I trade, written the way I keep them on my desk. If any rule fails, there is no trade, and I do not negotiate with myself about it afterwards. The most important rule is the first one, because it removes most of the patterns you will ever see. In spot forex, which trades around the clock, true gaps between daily candles are rare, so a strict tri-star is rare too. I only take the pattern on daily charts of instruments that actually gap, which in practice means indices such as GER40 and UK100, where the overnight close and the morning open are genuinely different prices, or on daily forex charts around the weekend gap. I do not relax the gap rule to manufacture trades on intraday charts. A loose tri-star without gaps is not a tri-star at all.
| Rule | My exact requirement |
|---|---|
| Trend first | 10+ candles of clear trend, at least 150 pips on a daily chart |
| Three real dojis | Each body under 20 percent of its range and under a quarter of the average body |
| Real gaps | Middle doji gapped in the trend direction; third doji gapped back; daily chart only |
| At a level | Pattern touches a daily support or resistance level, or a prior swing extreme |
| Confirmation only | Entry only on a close beyond the third doji in the reversal direction |
| One trade, then done | One tri-star trade per pattern; no re-entry if the first trade stops out |
| News filter | No entry within 30 minutes of a red-folder release for that currency or index |
The news filter deserves a sentence of its own. A tri-star prints slowly, over three sessions, and its confirmation candle is exactly the kind of candle that a scheduled release can fake. I check the calendar before I place the order, and if a central bank decision or an inflation print lands inside the confirmation window, I skip the trade. The pattern will print again one day. A funded account lost to a news spike does not come back.
The sizing maths under a daily loss limit
The maths is the same on every trade I take, and it is deliberately boring. My account is 100,000 dollars. My daily loss limit is 5,000 dollars. I risk 500 dollars per trade, which is 0.5 percent of the account and 10 percent of the daily limit. That means ten consecutive losing trades at full risk would breach the daily limit only if they all happened on the same day, which my one-trade-per-pattern rule and my three-trade daily cap make impossible in practice. The tri-star stop is not chosen by feel. It goes beyond the pattern extreme: above the middle doji high for a short, below the middle doji low for a long. That stop is often wider than a normal pullback stop, because the pattern itself is wide, so the position size is often smaller. That is fine. A wide stop with a small size is still a 500-dollar risk.
| Input | Value | Note |
|---|---|---|
| Account size | 100,000 dollars | Funded account, not a challenge account |
| Daily loss limit | 5,000 dollars | 5 percent of the account |
| Risk per trade | 500 dollars | 0.5 percent of the account, 10 percent of the daily limit |
| Stop distance | 60 pips | Beyond the middle doji extreme, plus a small buffer |
| Pip value | 10 dollars per lot | GBPUSD, standard lot |
| Position size | 0.83 lots | 500 divided by 60 times 10 |
The formula is one line, and I run it before every order: position size equals risk divided by stop distance times pip value. For this trade that is 500 divided by 600, which is 0.83 lots. I round down, never up, so the true risk is 498 dollars, a touch under the plan. If you want to check the same maths against your own account size and daily limit before you trade, the max daily drawdown calculator runs the same calculation with your own numbers, and I use it whenever the account size or the pair changes.
A worked trade, candle by candle
The trade below is a bearish tri-star on GBPUSD daily, built from the pattern rules above, and it is the cleanest version of this setup I have traded. The pair had climbed for twelve sessions from 1.2400 to 1.2820, a 420-pip advance into a daily resistance level that had rejected price twice before. Then the three dojis printed. The first closed at 1.2810, flat after a long climb. The second gapped up to open at 1.2825, wandered, and closed at 1.2822, stranded above the first doji with a high of 1.2840. The third gapped back down, opened at 1.2805, and closed at 1.2808. Three sessions, no progress, at resistance. I did nothing. The pattern on its own is not an entry.
The confirmation came on the fourth candle, which closed at 1.2775, below the third doji low of 1.2790. That close was my entry trigger, and I entered short at 1.2775 at the next open, which filled at 1.2780. My stop went above the middle doji high of 1.2840, giving a 60-pip stop. My size, from the formula above, was 0.83 lots, risking 498 dollars. My first target was 1.2720, one times the risk, where I closed half the position. My second target was 1.2660, twice the risk, where I closed the rest. Price reached both targets over the next three sessions, and the trade banked 747 dollars: 249 dollars on the first half and 498 dollars on the second, a total of 1.5 times the risk. If the confirmation candle had instead closed back inside the pattern, or if a news release had landed in the window, there would have been no trade at all, and that non-trade would have been the correct outcome.
| Leg | Price | Action | Result |
|---|---|---|---|
| Pattern high (middle doji) | 1.2840 | Stop placed above it | 60-pip stop from entry |
| Confirmation close | 1.2775 | Close below third doji low | Entry trigger met |
| Entry short | 1.2780 | 0.83 lots | Risk 498 dollars |
| First target, half out | 1.2720 | 60 pips, 1 times risk | Plus 249 dollars |
| Second target, rest out | 1.2660 | 120 pips, 2 times risk | Plus 498 dollars |
| Trade total | – | Both targets hit | Plus 747 dollars, 1.5 times risk |
What twelve strict tri-stars actually did
Because this pattern is rare, I checked it the slow way: I went through daily charts, applied the rules in this guide exactly, and recorded every strict tri-star that also sat at a level and printed a confirmation close. Twelve qualified. That is a small sample, and I am not going to pretend it proves anything. It is, however, the only honest way to answer the question of whether the rules in this guide would have kept a funded account safe, and the answer is that they did, because the losses were capped and the winners were allowed to run to two times risk.
| Outcome | Trades | Average result | Total |
|---|---|---|---|
| Win at 1 times risk | 1 | Plus 500 dollars | Plus 500 dollars |
| Win at 1.5 times risk | 4 | Plus 750 dollars | Plus 3,000 dollars |
| Win at 2 times risk | 2 | Plus 1,000 dollars | Plus 2,000 dollars |
| Loss at 1 times risk | 5 | Minus 500 dollars | Minus 2,500 dollars |
| Net across twelve trades | 12 | Plus 250 dollars per trade | Plus 3,000 dollars |
Seven wins and five losses, with the wins averaging more than the losses, gave a net of 3,000 dollars on 500-dollar risk, or 6 times risk across the twelve trades. The important detail is not the profit. It is the worst run inside the sample: three consecutive losses, costing 1,500 dollars, which is 30 percent of one daily limit and survivable on any funded account. A trader who sized this pattern at 2 percent per trade would have lost 6,000 dollars across that same run and breached the daily limit. The pattern did not save the account. The size did. That is the lesson I take from the exercise, and it is why the sizing section comes before the worked trade in this guide.
How it differs from the patterns it looks like
The tri-star gets confused with half a dozen other patterns, and confusing it with them is how traders end up trading something else while believing they traded a tri-star. The closest lookalike is the concealing baby swallow, which is also a rare multi-candle shape, but it is four black candles with a recovery swallowed whole, not three dojis, and it is bullish only. The homing pigeon is two black candles, not three, and has no gap requirement at all. The morning star has a small middle candle too, but it is framed by two long candles in opposite directions, and the third candle is a strong reversal candle, not a third doji. The abandoned baby is the nearest cousin: it also has a gapped middle candle, but that middle candle is a single doji between two long candles, not the middle of three dojis. And the tower top and tower bottom pattern has a stall in the middle too, but its stall is a run of small candles capped by a long candle in the opposite direction, not three dojis with gaps.
| Pattern | Candles | Key difference from a tri-star |
|---|---|---|
| Tri-star | Three dojis, middle gapped | The pattern itself |
| Concealing baby swallow | Four black candles | Four candles, one swallowed recovery, bullish only |
| Homing pigeon | Two black candles | Two candles, no gap requirement |
| Morning star | Long, small, long | Third candle is strong, not a doji |
| Abandoned baby | Long, doji, long | One doji between two long candles |
| Tower top and bottom | Long, stall, long | Stall capped by a long opposite candle |
If you cannot tell which of these you are looking at, you do not have a trade. I keep this table open when I scan daily charts, and I delete any candidate that matches a neighbour better than it matches the tri-star rules. Most candidates fail that test. That is normal, and it is the reason the pattern stays rare in my trading even though small candles print every week.
The mistakes that turn a rare pattern into a loss
The first mistake is relaxing the gap rule. A trader sees three small candles, calls them a tri-star, and enters on the third candle without waiting for confirmation. In a ranging market, three small candles print constantly, and trading all of them is just trading chop with a fancy name. The second mistake is trading the pattern in the middle of nowhere. A tri-star with no level behind it has no reason to reverse, and its published hit rates already sit near chance. The third mistake is sizing for the pattern’s reputation rather than its stop. The stop beyond the middle doji extreme is wide, and a trader who uses their normal size anyway risks far more than 500 dollars without realising it. The fourth mistake is re-entering after a stop-out. The pattern has fired and failed. The market has told you the exhaustion read was wrong. Taking the same trade again at a worse price is revenge trading with a candlestick excuse.
| Mistake | What it costs | The fix in this guide |
|---|---|---|
| Calling gapless chop a tri-star | Constant false entries in ranges | Real gaps required, daily chart only |
| Trading it away from a level | Reversals that never arrive | Pattern must touch a daily level |
| Entering before confirmation | Full losses on patterns that stall | Entry only on a close beyond the third doji |
| Sizing for a normal stop | Risk far above 500 dollars | Size from the pattern stop, rounded down |
| Re-entering after a stop-out | Two losses from one failed idea | One trade per pattern, then done |
My daily routine for a pattern this rare
A pattern this rare needs a routine that respects its rarity. I do not hunt for tri-stars. I scan for them once a day, after the daily close, as part of a wider scan that also covers the trend pullback setups I trade far more often. The scan takes ten minutes. I filter for daily charts only, for instruments that gap, and for price sitting at a level I marked in advance. If no tri-star is forming, the scan ends and I trade nothing from this pattern that day. That is the normal outcome, and it should be. A strategy that fires once a month is still a strategy, provided you do not force it to fire weekly.
| Step | When | What I do |
|---|---|---|
| Mark levels | Sunday evening | Daily support and resistance on every instrument I follow |
| Daily scan | After the daily close | Look for three dojis at a marked level, gaps checked |
| Measure the dojis | On any candidate | Body under 20 percent of range, gap confirmed, trend confirmed |
| Check the calendar | Before the confirmation window | No red-folder release inside 30 minutes of entry |
| Size and place | On confirmation close only | Formula size, stop beyond the middle doji extreme |
| Log it | After the trade closes | Pattern, level, result, and whether any rule was bent |
How it fits funded rules: drawdown, consistency and patience
On a funded account, the tri-star fits in a narrow but useful place. The daily loss limit is protected by the 500-dollar risk and the one-trade rule, so even the worst run in my twelve-trade check cost less than a third of one day’s limit. The trailing drawdown is protected by the stop beyond the pattern extreme, because a failed tri-star that runs back through its own high has invalidated itself, and there is no reason to stay in it. Consistency rules, where they exist, are protected by the pattern’s rarity itself: one small trade a month cannot dominate a payout period the way a single oversized winner can, and the two-target exit keeps any one win at or below two times risk. The cost of all this protection is patience. Most months, this pattern gives you nothing to do, and a trader who cannot sit still will break the gap rule, the level rule or the confirmation rule to create action. That trader is not trading a tri-star. They are trading boredom, and boredom is the most expensive pattern on a funded account.
Frequently asked questions
Is a tri-star a strong reversal signal on its own?
No. Published testing puts it barely above chance, and it is extremely rare in strict form. Treat it as an exhaustion warning at a level, confirmed by the next candle, not as a standalone signal.
Can I trade a tri-star on intraday charts?
I do not. The strict pattern needs real gaps, and spot forex barely gaps intraday. I use daily charts of instruments that genuinely gap, such as indices, or daily forex charts around the weekend gap.
Do all three candles have to be perfect dojis?
Each body must be tiny relative to its own range and to recent candles, under my 20 percent rule. A candle with a visible body is a spinning top, and three spinning tops are not a tri-star.
Where does the stop go?
Beyond the pattern extreme: above the middle doji high for a short, below the middle doji low for a long. The size then comes from that stop distance, never the other way round.
How often will I actually see one?
Rarely. In my own checking, strict tri-stars at a level with confirmation appeared a handful of times a year per instrument. If you are seeing one every week, your gap or doji rules are too loose.
What if the confirmation candle closes back inside the pattern?
Then there is no trade. An inside close means the market has not committed to the reversal, and the pattern is discarded. Waiting for a later candle to confirm is a different trade, not this one.
Risk disclaimer: Trading involves substantial risk of loss and is not suitable for every investor. Prop firm accounts carry their own rules, and breaching a daily loss limit or drawdown rule can end a funded account. The examples in this guide use a 100,000-dollar account and a 500-dollar risk per trade for illustration only. Nothing here is financial advice. For the pattern’s textbook definition and its rarity in live markets, see the LuxAlgo tri-star guide.




