Guides / Cookbooks

Cookbooks · Risk and execution · Intermediate · about 8 min

Size positions by volatility regime

Use ATR and realized-volatility rules to place each symbol in a calm, normal or high regime, and scale position size to match.

1. When to use it

   The same dollar size is a very different risk in an index ETF and in a high-beta chip stock. Rules that measure the current regime give a size multiplier before any forecast is involved.

2. The request

   atr_pct(14) is the 14-day average true range as a percentage of the close. hv_high is asked in English; the engine compiles it to a realized-volatility rule.

   {
     "model": "ora-fin1",
     "state": {
       "symbol": "NVDA"
     },
     "questions": {
       "atr_high": {
         "type": "yes_no",
         "ask": "atr_pct(14) > 3"
       },
       "atr_mid": {
         "type": "yes_no",
         "ask": "atr_pct(14) > 1.5"
       },
       "hv_high": {
         "type": "yes_no",
         "instructions": "Is its 20-day historical volatility above 40%?"
       },
       "drop5": {
         "type": "yes_no",
         "instructions": "Will it drop at least 5% at some point in the next month?"
       }
     }
   }

   Open this in the Playground

3. The real response

   NVDA:

   {
     "model": "ora-fin1",
     "answers": {
       "atr_high": {
         "type": "yes_no",
         "yes_no": 0.5548,
         "rule": "atr_pct(14) > 3",
         "basis": "rule"
       },
       "atr_mid": {
         "type": "yes_no",
         "yes_no": 0.9463,
         "rule": "atr_pct(14) > 1.5",
         "basis": "rule"
       },
       "hv_high": {
         "type": "yes_no",
         "yes_no": 0.1582,
         "rule": "std(log(C / C[-1]), 20) * sqrt(20 / 19) * sqrt(252) > 0.4",
         "basis": "pattern"
       },
       "drop5": {
         "type": "yes_no",
         "yes_no": 0.523
       }
     },
     "usage": {
       "input_tokens": 395,
       "output_tokens": 16
     },
     "context": {
       "source": "server",
       "symbol": "NVDA",
       "as_of": "2026-10-30"
     },
     "version": "ora-fin-0.1.1"
   }

   The English question became std(log(C / C[-1]), 20) * sqrt(20 / 19) * sqrt(252) > 0.4 (annualized 20-day volatility over 40%). Across three symbols:

   Symbol	atr_pct(14) > 1.5	atr_pct(14) > 3	20-day HV > 40%	Multiplier
   SPY	0.0537	0.0537	0.0537	1×
   NVDA	0.9463	0.5548	0.1582	0.75×
   AMD	0.9463	0.9441	0.938	0.5×

   NVDA's atr_high of 0.5548 is just past the 3% line: ATR is a touch above 3%, not far above it. A belief that close to 0.5 is the engine saying "borderline".

4. The code

   cURLPythonJavaScriptPHPGoJavaC#RubyCopy

   Requires curl 7.76+. Save as run.sh, then: bash run.sh

   curl -sS https://oramodel.com/v1/systemone \
     -H "Authorization: Bearer $JEV_API_KEY" \
     -H "Content-Type: application/json" \
     -d '{
     "model": "ora-fin1",
     "state": {
       "symbol": "NVDA"
     },
     "questions": {
       "atr_high": {
         "type": "yes_no",
         "ask": "atr_pct(14) > 3"
       },
       "atr_mid": {
         "type": "yes_no",
         "ask": "atr_pct(14) > 1.5"
       },
       "hv_high": {
         "type": "yes_no",
         "instructions": "Is its 20-day historical volatility above 40%?"
       },
       "drop5": {
         "type": "yes_no",
         "instructions": "Will it drop at least 5% at some point in the next month?"
       }
     }
   }'

   The multiplier in the table comes from this function:

   def size_multiplier(a):
       if a["atr_high"] >= 0.6 or a["hv_high"] >= 0.6:
           return 0.5      # high regime
       if a["atr_mid"] >= 0.6:
           return 0.75     # normal to busy
       return 1.0          # calm
   
   a = {k: v["yes_no"] for k, v in data["answers"].items()}
   shares = int(base_shares * size_multiplier(a))

5. Act on it

   Apply the multiplier before the order is built, and log the regime with the trade. Re-check after each close: regimes change faster than most position-sizing tables are updated.

6. Variations

   • Volatility targeting: size so that shares × ATR is a fixed dollar amount (read ATR from your own data and use the rule only as the regime label).
   • Use the beliefs smoothly instead of steps: multiplier = 1 − 0.5 × atr_high.
   • Add drop5 to cut size further when the drawdown odds are high (Risk guardrail before an order).

Information, not investment advice. Every number on this page is a real response for the session shown; answers change with each session.

What's next

• Risk guardrail before an order