Buy Vs Build: Using ThinkOrSwim to Kill subscriptions

[Randolph_Nomad]

Good Day all,
I've been a member for a few years, and everytime i use something on this site it teaches me; improving my technical analysis AND learning to read signals.

I decided to give some coding a try. forgive the pseudo program mgr context, but its easier for me to manage the approach this way.

Situation: there are services in the market (MTOptions and OptionRun) that chareg approx $140/mth to provide option signal trades. their records are pretty good, but the method doesn;t appear to be significantly proprietary. it seems to be a blend of the Stochastic forecasting, mean reversion & momentum components, with predictive signals and time based thresholds.
Mission/Goal: Use TOS to duplicate this process, Thus reducing or eliminating 3rd party need to generate cash consistently.
Approach:

• Build scanner Stochastic Processes
  • Capture Randomness & Probability Weightings for Directional Moves
• Include Momentum & Mean-Reversion Components
• Calibrate the Stochastic Process
• Generate Probability Readings
• Identify Attractive Option Strikes & Expirations
  • Use one of the 3rd party vendors as a baseline (Parallel test if the toll comes up with similar or identical recommendation)
• Implement Entry & Exit Criteria
  • Duplicate criteria from 3rd party vendor
• Combine with Technical & Fundamental Analysis
  • Identify what additional analysis/existing charts can improve confidence in execution of options trade

Execution (in progress): Since TOS cannot perform advanced stochastic differential equations, the approach must be hybrid. leveraging or building TOS studies, using momentum oscillators, mean reversion signals, IV data, and basic probability estimations.
My first step was to try and build a Stochastic Parameter:

Goal: Approximate a stochastic process that:

• Tracks near-term volatility dynamics (implied or realized).
• Flags momentum vs. mean-reversion conditions.
• Outputs a “probability reading” (or score) indicating the likelihood of a near-term run.

here's the code for the study: as I continue, any feedback is greatly appreciated

declare lower;

# -- 1) Declare inputs as "constant" values for lengths --
input HVLength = 20;       # Historical Volatility lookback
input RSIlength = 14;      # RSI lookback
input BollLength = 20;     # Bollinger (mean-reversion) lookback

# -- 2) Basic Price & Volatility Inputs --
def price = close;

# For HV, we can approximate with the standard deviation of log returns:
def dailyReturn = Log(price / price[1]);
def HV = StDev(dailyReturn, HVLength);
def IV = imp_volatility();  # TOS built-in implied volatility for the underlying

# -- 3) Momentum (RSI) --
# Instead of RSI(price, rsiLen), use RSI(length = RSIlength)
def rsiValue = RSI(length = RSIlength);

# -- 4) Mean-Reversion (Bollinger Z-Score) --
def avgPrice = Average(price, BollLength);
def stDevPrice = StDev(price, BollLength);
def zScore =
    if stDevPrice != 0 then
        (price - avgPrice) / stDevPrice
    else
        0;

# -- 5) Composite Probability-Like Score --
# Momentum Part: normalized RSI
def momentumPart = (rsiValue - 50) / 50;  # -1 (oversold) to +1 (overbought) scale

# Volatility Part: comparing IV and HV
def volPart =
    if IV < HV then
        1  # “Cheap” implied volatility vs realized
    else
        HV / IV;  # If IV > HV, ratio < 1

# Mean-Reversion Part: if zScore < -1, bullish snapback, if > +1, bearish
def mrPart =
    if zScore < -1 then
        1
    else if zScore > 1 then
        -1
    else
        0;

# Weight them (adjust coefficients to taste)
def stochParamRaw = (0.5 * momentumPart) + (0.3 * volPart) + (0.2 * mrPart);

# Convert raw value to 0-100 scale, just for a nicer read
def stochParamScore = Round(50 + 50 * stochParamRaw, 0);

# -- 6) Plot the "StochasticParam" Score --
plot SPScore = stochParamScore;
SPScore.SetDefaultColor(Color.CYAN);