This advice is everywhere now:
Stack ten indicators, wait for three of five to agree, then size up by how many line up.
THIS IS TOTAL BS.
I run 5+ live strategies and not one of them uses more than two signals.
Those bots already printed me 6 figs.
PUBLICLY.
The freshest example:
https://polymarket.com/@PMTraderAdam
Standard sweeper bot.
Already hit $26k in two months on sweeping 0.0001 cents.
As you can see, the best ones are stupidly simple.
So this is the honest version of the indicator guide everyone shares.
I'm going to walk all ten, fast.
Then I'm gonna cut eight of them, and show you the only two that tell you something your price chart doesn't already.
Stick with me cause the cut is the whole point.
First, why stacking 10 signals is a TRAP
Here's what nobody selling the ten-indicator stack will admit:
Most of those indicators are built from the same price you're already looking at.
RSI, MACD, EMA, VWAP, pivots — all of them are just past price, transformed and smoothed.
Adding five flavors of the same number is not five signals.
It's one signal wearing five hats and lagging more with every hat.
Every extra condition you bolt on makes your bot more fragile, not stronger.
You're not adding edge.
You're carving a tighter and tighter pocket that happens to fit last week's noise.
That's the definition of overfitting.
And it's exactly why the ten indicator backtest shows 80% and the live bot bleeds.
There's a simple test for this.
Split your data into three folds.
Train on two, test on the third, then rotate.
If your beautiful stack only prints on one slice and dies on the others, it was never an edge.
It was a story you told yourself about random noise.
The strategies that survive a three-fold test almost always use one or two signals.
Not ten.
10 indicators explained
1. RSI
A momentum oscillator built entirely from past price.
It tells you price moved fast, which you can already see on the chart.
It also stays "overbought" for hours, so it fakes you out constantly on a 15m window.
Verdict: CUT
Lagged price in a fancy wrapper.
2. MACD
Two moving averages and the difference between them.
Useful as a slow trend description, useless as a 15m trigger because it lags by design.
By the time the histogram confirms, the move you wanted is half over.
Verdict: CUT
Too slow for a 15-minute window.
3. VWAP
Volume-weighted average price big money benchmarks against.
It's a real level and price does revert to it, but it's still a backward-looking average.
Fine as a glance, not a signal you bet on alone.
Verdict: CUT as a trigger
A level to watch, nothing more.
4. EMA 9 / 21
Faster moving averages, same problem as MACD.
A crossover is just a smoothed restatement of a move that already happened.
Verdict: CUT
It's MACD with fewer steps.
5. Pivot points (HL10)
The recent high and low, genuine memory levels where buyers and sellers showed up before.
This one is closer to useful, because levels are real.
But on its own it's just lines on a chart with no order flow behind them.
Verdict: CUT as a standalone
Useful only as a level to watch.
6. Funding rate
Now we're leaving the price chart, which is the right direction.
Funding tells you how leaned the crowd is and extreme positioning is reversal fuel.
But it updates every 8 hours, so it's macro context, not a 15-minute trigger.
Verdict: CONTEXT
A size filter at best.
7. CVD (cumulative volume delta)
This one carries real information your price chart does not.
We'll come back to it in depth, because it's a keeper.
Verdict: KEEP
8. Liquidation heatmap
A map of where leveraged positions get force-closed, which acts like a magnet for price.
Genuinely useful as a target predictor, but not a clean entry trigger, and it's noisy.
Verdict: CONTEXT
Good for where price is headed, bad for do I enter now.
9. OBI (order book imbalance)
The other keeper.
It reads the actual resting liquidity, not a transformation of past price.
We'll come back to this one too.
Verdict: KEEP
10. Open interest
How much money is actually in position right now.
Like funding, it's real positioning data, but it's confirmation and context, not a 15m trigger.
Verdict: CONTEXT
Pairs with funding.
MUST HAVE indicators
Notice what survived.
Not a single price-derived indicator.
The two keepers both read order flow, the one thing your candles cannot show you.
CVD — what the aggressive traders are actually doing
CVD is the running total of market buys minus market sells.
It answers one question: who is hitting the book harder, buyers or sellers?
This is information price alone hides from you.
Price can sit dead flat while CVD quietly climbs, which means buyers are eating every offer with no breakout yet.
That's silent accumulation and CVD sees it before RSI or MACD ever will.
The strongest single use is divergence.
- Price prints a new high but CVD prints a lower high → the move has no real buying behind it and you fade it.
- Price prints a new low but CVD makes a higher low → sellers are exhausted and you fade the low.
CVD catches reversals a full move before the lagging stuff confirms them.
OBI — what's actually resting on the book right now
OBI is the ratio of resting bid size to resting ask size near the top of the book.
A big stack of bids and thin asks means the next aggressive seller gets eaten and price lifts.
The reverse pushes price down.
This is the only indicator on the whole list that updates in real time and tells you about pressure that hasn't happened yet.
It moves in seconds, so you use it as the final go or no-go in the last moments before entry.
Everything else on the list describes the past. OBI describes the immediate next move.
How to actually use just 2
Your entire stack is this:
- CVD tells you the real direction underneath the price.
- OBI tells you if the book agrees right now.
When both point the same way, you have a reason to enter that isn't just re-chewed price.
When they disagree, you skip.
That's it.
No five-of-ten confidence score, no stack of lagging oscillators all voting on the same candle.
Two signals that carry actual information, and the discipline to skip when they don't line up.
If you want context, glance at funding and open interest for crowd positioning, and the liquidation map for where price is being pulled.
But context sizes your bet, it does not make it.
The trigger is two signals, never ten.
CRUCIAL PART
Backtest your strategy before going live.
This is how most devs lose EVERYTHING trying to launch asap.
One tool lets you simulate everything with no money lost.
Just test your bot here first and go live once it shows good performance.
If not, keep developing.
Come back ONLY after it prints consistently.
Fair conclusion
The ten-indicator stack isn't sophisticated.
It's a confidence trick you play on yourself, and the backtest plays along, because it overfits to whatever you feed it.
Two real signals that survive a three-fold test will beat ten lagging ones every single time.
Cut the eight that are just price in a costume.
Keep the two that read the order flow.
Then put your real energy where the edge actually lives (execution, latency, and not getting front-run), which is a whole different guide.
You can find it in articles section on my profile.
And if you wanna connect, join my small TG chan.
Link: https://github.com/PMTraderAdam/500-per-day-polymarket-bot
Here we discuss everything together and help each other.
Good luck and see you there.
Top comments (0)