Adventures of a Single Dad

Independent quant researcher. Production-grade tools, walk-forward validated systems, open-source Python. Single dad grinding 2–5 a.m. to ship edge.

[Adventures of a Single Dad @LDUnbreakable]

i like your work. Whenever I'm done with this massive project, we should partner on your solar system market project. It was an obsession of mine early on. Have mountains of files, books, research papers, indicators, etc. I could run it through claude for optimizations. While I'm unsure the outcome, it's like the little thing in back of my mind that was never finished.

[Adventures of a Single Dad @LDUnbreakable]

I am excited about that. My read on it is that it's the most important thing the factory has made. Version 3 will look in 2 Monte Carlo runs per bar - middle and last 10%(whatever is needed to complete the run). Symbols are so volatile that it's active to make a trade decision. Can't wait to explore this. It's real edge.

[Adventures of a Single Dad @LDUnbreakable]

🚀 Just shipped the FIRST Dynamic Risk Assessment v2 for ThinkOrSwim.

Real Monte Carlo P(SL) + embedded HMM 8-regime engine + 10-mode adaptive Laguerre RSI (Lyapunov & Orbital gamma included) + self-learning feedback that actually tunes stops, targets & confluence live.

No more static ATR guesses. Every bar delivers:
• Probabilistic risk score
• Regime-adjusted position sizing
• Adaptive stop & target levels
• Built-in performance analytics

/ES 5-min example below. This is next-level for TOS.
These indicators will soon be available exclusively to MarketFragments subscribers via secure per-trader plugins (full power + full IP protection).

https://t.co/Z3zpKA6LzF

[Adventures of a Single Dad @LDUnbreakable]

@NovumMarkets I saw the market regime stuff. I too work heavily with market regimes as a foundation to trade indicators and dynamic risk analysis. Ever want to bounce ideas back and forth, hit me up. mi casa su casa sort of thing

[Adventures of a Single Dad @LDUnbreakable]

Labels are a little difficult to read. Here is the top label :

[Adventures of a Single Dad @LDUnbreakable]

@NovumMarkets cool pic

[Adventures of a Single Dad @LDUnbreakable]

Not that I'm the 'cat's meow' with Markov Chains but agree a lot with the article. Specifically, "X Cannot Do * Overcome the efficient market hypothesis without an extra source of edge". However, from my research in creating a strategy factory, I use an 8 state market regime transitions and apply specific trade plans to each category and then race to the finish when pairing an indicator. While 3 regime occurrences dominate with my model, sub classifications of the 3 have been profitable in back test. Still working it. I love this stuff. One day, I will be super nerd creating novel ways to trade. Hopefully sooner than later :)

[Adventures of a Single Dad @LDUnbreakable]

here you go #grok # ─────────────────────────────────────────────
# │ https://t.co/AcXUSORuoC | DNA & Market │
# │ [email protected] │
# │ https://t.co/Z3zpKA6LzF │
# ─────────────────────────────────────────────
# Time →
# │
# █ █ █│ █
# █ █ █ │ █ █
# █ █ █ │ █ █ █ ╭─╮
# █ █ █ │ █ █ █ █ ╭─╯ ╰─╮
# █ █ █ │ █ █ █ █ █ █ ╭─╯ ╰─╮
# █ █ █ │ █ █ █ █ █ █ █ █ ╭─╯ ╰─╮
# █ █ █ │ █ █ █ █ █ █ █ █ █ █╭─╯ ╰─╮
# █ █ █ │ █ █ █ █ █ █ █ █ ╰─╮ ╭─╯
# █ █ █ │ █ █ █ █ █ █ ╰─╮ ╭─╯
# █ █ █ │ █ █ █ █ ╰─╮ ╭─╯
# █ █ █ │ █ █ ╰─╮ ╭─╯
# █ █ █ │ █ ╰─────╯
# ──────┴──────────────────────────────────────────────────────────────
# T1 T2 T3 T4 T5 T6
#
# ═══════════════════════════════════════════════════════════════════════════
# https://t.co/AcXUSORuoC │ DNA & Market
# PIVOT ENGINE — Session-Aware Regression-Gated Swing Pivots
# [email protected] · https://t.co/Z3zpKA6LzF
# ═══════════════════════════════════════════════════════════════════════════
# Research preview. Selected from a 9-variant slope bake-off on 5yr /ES 5m data.
#
# WHAT IT DOES
# Marks confirmed swing pivots and draws their levels as support/resistance.
# Direction is decided by a true least-squares regression slope of price over
# `pivotLength` bars, filtered by an ATR "significance gate": a turn only
# registers when the slope is steep enough to clear the market's own noise.
#
# WHY REGRESSION + GATE (the short version of the research)
# - Across 9 slope kernels, directional hit-rate was ~76% no matter which —
# the kernel is a swing-SCALE dial, not an accuracy edge.
# - A slope-significance GATE is the real lever: it trims chop, so pivots get
# fewer, larger and cleaner, and the levels stop "wiggling" (repaint match
# went to 100% once the gate was on).
# - SESSION is the strongest conditioner found: RTH pivots ran ~2x the swing
# size and ~5pp higher success than overnight (ETH). So the gate is
# session-switched: TIGHT in RTH (trend/breakout), LOOSE in ETH (range/S-R).
#
# NON-REPAINTING: the plotted levels (HH/LL) update ONLY at a confirmed turn and
# snap to an extreme that already printed. They do not move once drawn — safe
# to anchor stops/breakouts to. (The bar-marker wedges are cosmetic.)
#
# HOW TO USE
# RTH (tight gate): treat levels as breakout/trend pivots — price tends to run
# through them, big swings. ETH (loose gate): treat levels as fade/S-R —
# overnight revisits and rejects them more often.
# ═══════════════════════════════════════════════════════════════════════════

declare upper;

# ─── INPUTS ─────────────────────────────────────────────────────────────────
input pivotLength = 13;
input sessionAware = yes; # auto-switch the gate by session
input gateK_RTH = 0.15; # TIGHT gate during RTH (trend / breakout pivots)
input gateK_ETH = 0.05; # LOOSE gate overnight (range / S-R pivots)
input gateK_manual = 0.10; # used when sessionAware = no
input rthStart = 0930; # RTH window (chart timezone) — adjust to your data
input rthEnd = 1600;

input showLevels = yes; # horizontal S/R lines at last confirmed HH/LL
input showSwingPoints = yes; # bubble each completed swing's size in points
input showPivotMarks = yes; # up/down wedges at confirmed turns
input shadeSession = yes; # tint background during RTH
input showStats = yes;

# ─── COMMON ─────────────────────────────────────────────────────────────────
def c = close; def h = high; def l = low; def o = open;
def bn = BarNumber(); def nan = Double.NaN; def tick = TickSize();
def atr = WildersAverage(TrueRange(h, l, c), pivotLength);

# ─── SESSION + SESSION-SWITCHED GATE ─────────────────────────────────────────
def rth = SecondsFromTime(rthStart) >= 0 and SecondsTillTime(rthEnd) > 0;
def gateK = if !sessionAware then gateK_manual
else if rth then gateK_RTH else gateK_ETH;

# ─── SLOPE ENGINE: OLS least-squares regression of the price LEVEL ───────────
# Returns the forward-time slope (>0 = ascending). Body-independent by design —
# subtracting a per-bar midpoint measured volatility, not direction, in testing.
script RegSlope {
input Value = high;
input length = 13;
def n = length;
def Sx = n * (n - 1) / 2;
def Sxx = (n - 1) * n * (2 * n - 1) / 6;
def Sy = fold iy = 0 to n with sy1 = 0 do sy1 + Value[iy];
def Sxy = fold ixy = 0 to n with sxy1 = 0 do sxy1 + ixy * Value[ixy];
def denom = n * Sxx - Sx * Sx;
def raw = if denom != 0 then (n * Sxy - Sx * Sy) / denom else 0;
plot slope = -raw; # forward-time: >0 = rising
}

def slopeH = RegSlope(high, pivotLength).slope;
def slopeL = RegSlope(low, pivotLength).slope;

# ─── ATR SIGNIFICANCE GATE (session-switched) ────────────────────────────────
def noiseFloor = gateK * atr;
def ascH = slopeH > 0 and AbsValue(slopeH) > noiseFloor;
def descH = slopeH < 0 and AbsValue(slopeH) > noiseFloor;
def ascL = slopeL > 0 and AbsValue(slopeL) > noiseFloor;
def descL = slopeL < 0 and AbsValue(slopeL) > noiseFloor;

# ─── PIVOT STATE MACHINE (confirmed, non-repainting staircase) ───────────────
def uTrend = if bn < pivotLength then 0 else ascH and ascL;
def dTrend = if bn < pivotLength then 0 else descH and descL;

def zzstate = if uTrend then 100 else if dTrend then -100 else zzstate[1];
def uTr = if zzstate == 100 and zzstate[1] == -100 then 1 else nan; # up-turn confirmed
def dTr = if zzstate == -100 and zzstate[1] == 100 then 1 else nan; # down-turn confirmed

def xh = if !IsNaN(uTr) then h else if h > xh[1] and zzstate == 100 then h else xh[1];
def xl = if !IsNaN(dTr) then l else if l < xl[1] and zzstate == -100 then l else xl[1];

def ishh = if !IsNaN(dTr) then xh else ishh[1]; # confirmed resistance (swing high)
def isll = if !IsNaN(uTr) then xl else isll[1]; # confirmed support (swing low)

# ─── LEVEL PLOTS (non-repainting horizontal S/R) ─────────────────────────────
plot Resistance = if showLevels then ishh else nan;
Resistance.SetPaintingStrategy(PaintingStrategy.HORIZONTAL);
Resistance.SetDefaultColor(https://t.co/FlERnG3YFt);
Resistance.SetLineWeight(2);

plot Support = if showLevels then isll else nan;
Support.SetPaintingStrategy(PaintingStrategy.HORIZONTAL);
Support.SetDefaultColor(https://t.co/wR4QoDRw3q);
Support.SetLineWeight(2);

# ─── PIVOT MARKERS ───────────────────────────────────────────────────────────
plot PivotHigh = if showPivotMarks and !IsNaN(dTr) then ishh + 2 * tick else nan;
PivotHigh.SetPaintingStrategy(PaintingStrategy.BOOLEAN_WEDGE_DOWN);
PivotHigh.SetDefaultColor(https://t.co/FlERnG3YFt); PivotHigh.SetLineWeight(3);

plot PivotLow = if showPivotMarks and !IsNaN(uTr) then isll - 2 * tick else nan;
PivotLow.SetPaintingStrategy(PaintingStrategy.BOOLEAN_WEDGE_UP);
PivotLow.SetDefaultColor(https://t.co/wR4QoDRw3q); PivotLow.SetLineWeight(3);

# ─── SWING POINTS (magnitude of each completed leg, in points) ───────────────
def swingPts = AbsValue(ishh - isll);
AddChartBubble(showSwingPoints and !IsNaN(dTr), ishh, Round(swingPts, 1) + " pt", https://t.co/FlERnG3YFt, yes);
AddChartBubble(showSwingPoints and !IsNaN(uTr), isll, Round(swingPts, 1) + " pt", https://t.co/wR4QoDRw3q, no);

# ─── SESSION SHADING ─────────────────────────────────────────────────────────
AddCloud(if shadeSession and rth then Double.POSITIVE_INFINITY else nan,
Double.NEGATIVE_INFINITY, CreateColor(20, 30, 45), CreateColor(20, 30, 45));

# ─── LIVE STATS (the same metrics used in the research) ──────────────────────
def tot_up = CompoundValue(1, if !IsNaN(dTr) then tot_up[1] + 1 else tot_up[1], 0);
def tot_dn = CompoundValue(1, if !IsNaN(uTr) then tot_dn[1] + 1 else tot_dn[1], 0);
def lastHigh = if !IsNaN(dTr) then h else lastHigh[1];
def lastLow = if !IsNaN(uTr) then l else lastLow[1];
def succ_up = CompoundValue(1, if !IsNaN(dTr) and !IsNaN(lastHigh[1]) and xh > lastHigh[1] then succ_up[1] + 1 else succ_up[1], 0);
def succ_dn = CompoundValue(1, if !IsNaN(uTr) and !IsNaN(lastLow[1]) and xl < lastLow[1] then succ_dn[1] + 1 else succ_dn[1], 0);
def totRot = tot_up + tot_dn;
def overall = if totRot > 0 then (succ_up + succ_dn) / totRot else 0;

def p_ishh = ishh[1]; def p_isll = isll[1];
def sumPts = CompoundValue(1,
if (!IsNaN(uTr) or !IsNaN(dTr)) and !IsNaN(ishh) and !IsNaN(isll) then sumPts[1] + swingPts else sumPts[1], 0);
def avgPts = if totRot > 0 then sumPts / totRot else 0;

AddLabel(showStats, "MarketFragments Pivot Engine", Color.WHITE);
AddLabel(showStats, "Session: " + (if rth then "RTH (trend/breakout)" else "ETH (range / S-R)") +
" Gate K: " + gateK, if rth then Color.CYAN else https://t.co/IvUFkmOqx0);
AddLabel(showStats, "Rotations: " + totRot, Color.LIGHT_GRAY);
AddLabel(showStats, "Swing success: " + AsPercent(overall),
if overall > 0.80 then https://t.co/wR4QoDRw3q else if overall > 0.75 then Color.YELLOW else https://t.co/IvUFkmOqx0);
AddLabel(showStats, "Avg swing: " + Round(avgPts, 1) + " pt", Color.LIGHT_GRAY);
AddLabel(showStats, "Slope: " + Round(slopeH, 4) + " Floor: " + Round(noiseFloor, 4),
if AbsValue(slopeH) > noiseFloor then https://t.co/wR4QoDRw3q else Color.GRAY);

[Adventures of a Single Dad @LDUnbreakable]

🚨 Non-Repainting Pivot Engine Just Dropped – Built for Real Traders & Quants

I spent months building, testing, and stress-testing a clean pivot detection system that actually works in live markets.

No repainting.
No curve-fitting hype.
Just solid math + 5 years of ES futures data.

Here’s the release:

For Traders (the practical stuff):
• Detects real swings on the fly with zero look-ahead bias
• Automatically switches behavior between RTH and ETH sessions (huge edge)
• Clean ThinkOrSwim indicator with live stats box – ready to drop in today
• 81%+ verified pivots in RTH, rock-solid rejection/revisit stats

If you trade futures, indices, or any liquid instrument and hate fake levels, this one’s for you.

For Quants (the nerdy details):
• Least-squares regression slope + ATR significance gate + session-conditioned K factor
• Full walk-forward validation on unadjusted 5-min ES continuous contract
• Honest LSTM companion study (spoiler: directional accuracy hovers ~50% – exactly what the math predicted)
• All code, CSV data, and reconciliation notes public

Transparency over marketing. Numbers over narratives.

Full Release (free to read):

1. Main Paper → https://t.co/Vmn7tKTu8i

2. LSTM Companion (the “negative” result we still published) → https://t.co/QuCUkUfs8L

3. ThinkScript Library + Download → https://t.co/mlUzIppCnY

Drop it in your chart and tell me what you see.
Feedback, edge cases, or ideas for the next version all welcome.

#Trading #Futures #Quant #ThinkOrSwim #PivotPoints

(If you’re on mobile and the links don’t click, copy-paste them — they’re live.)

What do you think — drop your first impression below 👇

[Adventures of a Single Dad @LDUnbreakable]

@grok mmm its unlocked. Lets see what your thoughts are

[Adventures of a Single Dad @LDUnbreakable]

@Blueprint_So9 thanks. I just post about technical indicators and market research. I mostly like market research topics - hopefully, it's a full-time thing very soon. Nerd in the basement creating novel market tools is my goal :)

[Adventures of a Single Dad @LDUnbreakable]

Quick free technical indicator for the divergence nerds out there:

Turns out classic divergence doesn’t work quite the way most of us have been taught (especially the bull/bear symmetry thing).

Dug into it on /ES 5m and built this Market Cap-Adjusted Quantity Index with 6 different detection methods + actual backtest stats baked in. Hard bear signals have been showing some real edge.

Wrote it up here: https://t.co/uZv8x1Lwmw

Full ThinkScript + all the validation labels in the library discussion: https://t.co/Gp84V4roSw

Would love to hear what you think if you throw it on a chart.

[Adventures of a Single Dad @LDUnbreakable]

@TradeQuantiX don't we all :)

[Adventures of a Single Dad @LDUnbreakable]

@MoonDevOnYT strong work. I love the display regardless

[Adventures of a Single Dad @LDUnbreakable]

@LeppyrdTrading I don’t have my glasses on but it looks like it took a long signal on a down candle. Doesn’t matter because it worked

[Adventures of a Single Dad @LDUnbreakable]

@RoundtableSpace 90 seconds is quick. I can backtest a short very window as a gate to decease some methods I use, but 90 seconds to backtest, OOC over fit test and forward seems too good to be true. I’m curious - What analysis methods are being used ?

[Adventures of a Single Dad @LDUnbreakable]

@MoonDevOnYT You probably have more bots that 20 + accounts combined. For your bots, what platform are you using and/or how do integrate with live trading?

[Adventures of a Single Dad @LDUnbreakable]

Just open-sourced the whole Risk Assessment Suite 🔥

8 custom ThinkOrSwim indicators I built because I was tired of having a “solid win rate” and still getting chopped up by path dependence and bad sizing.

Inside you get:

-Live GBM Position Sizer (tells you exactly how big to go on the next trade)
-Kelly + Monte Carlo equity sims (see the real drawdowns before you live them)
-Bell curves (one symbol vs 4-stream portfolio)
--Randomized & Average Parameters curves
Trade System Expectancy with streaks
-Fibonacci Risk/Reward auto-sizer

All of it. No paywall. No “join for the good stuff.” Just free.

Full breakdown + every script here → https://t.co/eKZQvZJPke

Load ‘em up, mess with your own numbers, and tell me which one hits different for you.

Stop guessing on risk.
#ThinkOrSwim #ThinkScript #Trading #RiskManagement
https://t.co/8liEufDHzj

[Adventures of a Single Dad @LDUnbreakable]

Great read. There is so much you can do with Markov Chains. Bake in a little Geometric Brownian Motion + Family, little fusion for synthetic data in your MC. Pretty dynamic risk analyzer. A lot more tuned to specific strategy than Kelly Criteria. Not knocking Kelly Fractional as it’s the Gold Standard, but a lot of benefits going a little deeper.