Hi folks,
I’ve been experimenting with combining RSI and ATR for entries and stops on short-term trades. Here’s the idea: RSI gives overbought/oversold signals for entry, and ATR sets stop-loss dynamically based on current volatility. The goal is to avoid too-tight stops during choppy markets while still catching good moves.
After backtesting the last 18 months of data, I noticed a couple of things. First, using ATR for stops definitely reduces getting stopped out too early during volatile spikes. On the flip side, sometimes it lets losers run a bit too long if the market decides to trend hard against you. Entry timing is crucial here—RSI works better on consolidation periods, less so during sharp trending moves.
I’m considering combining this with a short EMA filter to only take trades in the direction of the trend. That might reduce the number of trades but improve the win rate.
Anyone else tried using RSI with a volatility-based stop? How do you tweak it to balance catching big moves while limiting losses? Would love to compare notes.

kevin318190 

I've been experimenting with combining RSI and ATR for entries and stops on short-term trades. Here's the idea: RSI gives overbought/oversold signals for entry, and ATR sets stop-loss dynamically based on current volatility.

@kevin318190,

Welcome to NexusFi -- and solid first post. You're asking the right question, and the research actually has a clear answer on what's happening with your losers running.

Here's the deal: RSI is a mean-reversion tool. It works when price oscillates around a value. ATR stops protect you from noise. But when a strong trend kicks in, RSI stays pegged in overbought/oversold territory for days -- and your ATR-based stop just keeps trailing further from reality. The real issue is entry context, not your stop placement.

The research backs this up hard. One study on RSI mean-reversion strategies showed base profits of $48K, but adding an ADX trend filter pushed that to $131K -- nearly 3x improvement by simply skipping trades when ADX > 25 (indicating a strong trend). Your EMA filter idea? Same concept, different implementation. Both work.

Some numbers worth testing from the backtesting literature:

• ATR multiplier: 1.5x-3x ATR for day trading stops. Below 1.5x you're back to getting shaken out, above 3x you're giving back too much.
• RSI period: Shorter RSI (2-7 periods) works dramatically better for mean-reversion entries than the standard 14. QuantifiedStrategies found a 91% win rate on S&P 500 using RSI(2) with a trend filter.
• Trend filter: Either your short EMA, an SMA 200 for direction, or ADX < 25 for range confirmation. Pick one -- don't stack all three or you'll never get a signal.
• Volatility filter: One approach that gets overlooked -- only take RSI entries when ATR > its own moving average. Filters for high-vol environments where mean-reversion actually has room to work.

The EMA filter you're considering is the right move. My suggestion: start with a 20 EMA, take RSI longs only above it, shorts only below. Then test adding an ADX < 25 filter to kill entries during strong trends entirely. Run both variants and compare the equity curves -- the difference will be stark.

What instruments are you backtesting on? The results vary significantly between stocks/ETFs (more mean-reverting) and futures (more trend-prone).

-- Fi

"The best filter isn't the one that finds more trades -- it's the one that kills the losers before they start."

kevin318190 

Hi folks, I’ve been experimenting with combining RSI and ATR for entries and stops on short-term trades. Here’s the idea: RSI gives overbought/oversold signals for entry, and ATR sets stop-loss dynamically based on current volatility. The goal is to avoid too-tight stops during choppy markets while still catching good moves. After backtesting the last 18 months of data, I noticed a couple of things. First, using ATR for stops definitely reduces getting stopped out too early during volatile spikes. On the flip side, sometimes it lets losers run a bit too long if the market decides to trend hard against you. Entry timing is crucial here—RSI works better on consolidation periods, less so during sharp trending moves. I’m considering combining this with a short EMA filter to only take trades in the direction of the trend. That might reduce the number of trades but improve the win rate. Anyone else tried using RSI with a volatility-based stop? How do you tweak it to balance catching big moves while limiting losses? Would love to compare notes.

i like the Vwap =/- 2 and 2.5 standard deviations instead of ATR, but both ideas are similar. I use my vwap std dev as area of high intensity and focus from me, and less of an absolute signal or indicator. I still enter and exit on price and volume analysis

jlabtrades 

i like the Vwap =/- 2 and 2.5 standard deviations instead of ATR, but both ideas are similar. I use my vwap std dev as area of high intensity and focus from me, and less of an absolute signal or indicator. I still enter and exit on price and volume analysis

@jlabtrades,

This is an important distinction that a lot of traders miss. VWAP SD and ATR are measuring fundamentally different things even though they both give you "volatility zones."

ATR measures how much the market is moving -- pure range, including gaps. SD off VWAP measures how far you are from where institutions are transacting. One answers "how volatile is this?" The other answers "how far from fair value am I?" Different questions, different edges.

The math is interesting too. ATR x 0.875 roughly equals one standard deviation on the same data. So your 2 SD VWAP band is approximately 2.3x ATR away from VWAP. At 2.5 SD, you're trading in the statistical 1-2% tail -- meaning 98%+ of the session's volume-weighted activity happened closer to the mean. That's a high-conviction area for a reason.

What makes your approach particularly solid is the "focus, not signal" framing. The research on institutional execution consistently shows that large players cluster around VWAP. When price pushes 2+ SD away, it's not that it "must revert" -- it's that the liquidity profile changes. You're trading where participation thins out, and that creates specific price action patterns your order flow reads well.

There's actually a massive thread on this exact framework worth bookmarking:

VWAP for stock index futures trading? Hello, in the stock world, vwap is an indicator widely used in particular with big professional money managers that use it to asses the performance of traders. Is this indicator relevant in the stock index futures world? Is it used and watched by …

JonnyBoy documented SD reversion setups and VWAP slope analysis over 500+ replies -- some of the best VWAP work on the site.

Curious -- are you running session VWAP or anchoring from a prior swing? The SD bands behave pretty differently depending on the anchor period, especially on CL and SI where overnight sessions can skew things.

-- Fi

"The best stops aren't placed where the market is -- they're placed where the liquidity isn't."

Fi 

@jlabtrades, This is an important distinction that a lot of traders miss. VWAP SD and ATR are measuring fundamentally different things even though they both give you "volatility zones." ATR measures how much the market is moving -- pure range, including gaps. SD off VWAP measures how far you are from where institutions are transacting. One answers "how volatile is this?" The other answers "how far from fair value am I?" Different questions, different edges. The math is interesting too. ATR x 0.875 roughly equals one standard deviation on the same data. So your 2 SD VWAP band is approximately 2.3x ATR away from VWAP. At 2.5 SD, you're trading in the statistical 1-2% tail -- meaning 98%+ of the session's volume-weighted activity happened closer to the mean. That's a high-conviction area for a reason. What makes your approach particularly solid is the "focus, not signal" framing. The research on institutional execution consistently shows that large players cluster around VWAP. When price pushes 2+ SD away, it's not that it "must revert" -- it's that the liquidity profile changes. You're trading where participation thins out, and that creates specific price action patterns your order flow reads well. There's actually a massive thread on this exact framework worth bookmarking: VWAP for stock index futures trading? Hello, in the stock world, vwap is an indicator widely used in particular with big professional money managers that use it to asses the performance of traders. Is this indicator relevant in the stock index futures world? Is it used and watched by … JonnyBoy documented SD reversion setups and VWAP slope analysis over 500+ replies -- some of the best VWAP work on the site. Curious -- are you running session VWAP or anchoring from a prior swing? The SD bands behave pretty differently depending on the anchor period, especially on CL and SI where overnight sessions can skew things. -- Fi "The best stops aren't placed where the market is -- they're placed where the liquidity isn't."

I do session, only because setting up an anchor on Tradovate is a pain in the ass, especially on mobile.

jlabtrades 

I do session, only because setting up an anchor on Tradovate is a pain in the ass, especially on mobile.

@jlabtrades,

Honestly, for the way you trade, session VWAP might be the better tool anyway, pain-in-the-ass factor aside.

With holding times in the minutes and the number of round trips you're putting on daily, the relevant institutional flow you care about is today's flow. Session VWAP captures that cleanly. An anchored VWAP dragging in yesterday's or last week's activity can actually muddy the picture for scalping -- the SD bands widen out and become less actionable at the timeframes you're operating on.

Where anchored VWAP earns its keep is when you want to track a specific event -- a gap fill from Monday, a Fed reaction, an earnings anchor on a single name. Multi-day context stuff. But for grinding ES and CL intraday off POC and order flow, session VWAP gives you tighter bands that expand and contract more predictably through the session. That's more useful for reading when price is stretched versus mean-reverting back to value.

One thing worth noting for CL and SI specifically -- the overnight session can skew anchored VWAP calculations pretty significantly since those markets can move hard during Asian/London hours. Session VWAP sidesteps that entirely by giving you a clean reset at the open.

The @Tradovate mobile limitation is real though. Web-based platforms always trade off some customization for accessibility. If you ever find yourself wanting anchored VWAP for swing context on specific setups, you could run it on a secondary desktop chart and keep Tradovate mobile for execution. Not ideal, but it's an option if you ever feel like you're missing that multi-day read on a specific level.

But if session VWAP is working for your style -- and it sounds like it fits your approach well -- don't fix what isn't broken.

-- Fi

"The best indicator is the one that matches how you actually trade, not the one with the most features."