Strategy vault / Machine learning
SP500 Hybrid Seasonal Trading Strategy
The S&P500 Hybrid Seasonal Trading Strategy is a quantitative strategy that trades stocks based on seasonal patterns. It combines an enhanced buy & hold system, technical indicator conditions, and volume flow…
Overview
The S&P500 Hybrid Seasonal Trading Strategy is a quantitative strategy that trades stocks based on seasonal patterns. It combines an enhanced buy & hold system, technical indicator conditions, and volume flow indicators to rotate between the better and worse performing months of the year.
Strategy Logic
The main trading signals and rules are:
- Go long at the open on the first trading day of October every year.
- If the VIX is above 60% or the 15-day ATR is above 90%, pause seasonal trading until volatility subsides later in the month or year.
- Exit longs/short at the open on the first trading day of August every year.
- Exit longs/short also if VIX exceeds 120% or if VFI crosses below -20 while its 10-day MA points down.
- Optional short selling enabled.
The strategy capitalizes on the uneven performance of the stock market through the year, going long during October-April which statistically outperformed and taking profit or shorting during the poorer performing months of May-September. Risk is also managed by pausing trading when volatility spikes, as measured by indicators like VIX and ATR.
Advantage Analysis
The S&P500 Hybrid Seasonal Trading Strategy has the following key advantages:
- Leverages established, stable seasonal patterns grounded in the empirically observed uneven monthly performance of the S&P500 index through the year.
- Incorporates multiple filter conditions like VIX, ATR and VFI to effectively filter out noise and generate more reliable trading signals.
- Configurable trading rules for going long/short and custom periods for seasonal entry and exit that facilitate testing and optimization.
- Embedded risk avoidance mechanisms via volatility measures like VIX and ATR thresholds to bypass effects from violent market swings.
- Supplementary signal input from volume flow indicator reflecting potential shifts in market participation.
Risk Analysis
Some potential risks include:
- Invalidation risk of historical patterns. Markets evolve stochastically so historical tendencies may not always endure.
- Missignal risk from technical indicators. VIX, ATR and VFI can also generate false signals.
- Suboptimal parameter risk. Further parameter testing and tuning is possible since current values may not be globally optimal.
- Additional shorting risks like unlimited losses.
Risks can be mitigated via more rigorous risk controls, combination of indicators, parameter tuning, machine learning etc.
Enhancement Opportunities
Possible optimization opportunities:
- Longer backtest periods for more training data.
- Introduce stop loss mechanisms to control loss per trade.
- Fine-tune parameters of indicators like VIX, ATR and VFI to find best combinations.
- Deploy machine learning models to enable adaptive optimization.
- Ensemble strategies to lower systemic market risk via non-correlation.
Conclusion
The S&P500 Hybrid Seasonal Trading Strategy synthesizes well established seasonal tendencies, technical timing indicators and money flow measures. By avoiding the worst performing months of the year and positioning in the seasonally stronger months supplemented by effective volatility gating, the framework can yield consistent alpha. The adaptable structure also provides useful modular components for practitioners to test, optimize and build upon. Additional data, stop losses, parameter tuning and ensembles present further opportunity to improve performance.
Parameters
| Name | Default | What it controls |
|---|---|---|
| v_input_1 | timestamp(2021-01-01) | (?Observation Window:)Start date: |
| v_input_bool_1 | false | timebooli_useSdate |
| v_input_2 | timestamp(2022-01-01) | End date: |
| v_input_bool_2 | false | timebooli_useEdate |
| v_input_int_1 | false | (?ig_ro)Lookback Shift: |
| v_input_bool_3 | true | (?ig_so)Long Only: |
| v_input_int_2 | 8 | Sell Month: |
| v_input_int_3 | 60 | Max VIX up: |
| v_input_int_4 | -20 | Critical VFI Sell: |
| v_input_float_1 | 1.5 | ATR/VIX Ratio: |
| v_input_3 | VIX | Volatility Index: |
| v_input_timeframe_1 | D | stringi_VItf |
| v_input_int_5 | 25 | Implied Volatility period: |
| v_input_int_6 | 15 | Historical Volatility period: |
| v_input_int_7 | 130 | VFI period: |
| v_input_int_8 | 10 | MA: |
| v_input_float_2 | 0.2 | VFI Coef Cuttoff: |
| v_input_float_3 | 2.5 | Volat.: |
| v_input_int_9 | 15 | ATR length: |
| v_input_bool_4 | false | (?ig_to)Show Table: |
| v_input_string_1 | 0 | Position:: position.middle_right | position.top_center | position.top_right | position.middle_left | position.middle_center | position.top_left | position.bottom_left | position.bottom_center | position.bottom_right |
| v_input_int_10 | false | Transparency: |
| v_input_color_1 | #000000 | Table Colors: |
| v_input_color_2 | #d6dae3 | colori_Tcrowe |
| v_input_color_3 | #cccccc | colori_Tcrowo |
| v_input_string_2 | 0 | Table Text:: size.small | size.huge | size.large | size.normal | size.auto | size.tiny |
| v_input_color_4 | #000000 | Text Colors: |
| v_input_color_5 | red | colori_Tcsigt |
| v_input_color_6 | navy | colori_Tctitt |
Source Pine Script
/*backtest
start: 2023-12-01 00:00:00
end: 2023-12-31 23:59:59
period: 1h
basePeriod: 15m
exchanges: [{"eid":"Futures_Binance","currency":"BTC_USDT"}]
*/
// TASC Issue: April 2022 - Vol. 40, Issue 4
// Article: Sell In May? Stock Market Seasonality
// Article By: Markos Katsanos
// Language: TradingView's Pine Script v5
// Provided By: PineCoders, for tradingview.com
//@version=5
strategy(title = "TASC 2022.04 S&P500 Hybrid Seasonal System",
shorttitle = "HSS v2.0",
overlay = true,
default_qty_type = strategy.percent_of_equity,
default_qty_value = 10,
initial_capital = 100000,
currency = currency.USD,
commission_type = strategy.commission.percent,
commission_value = 0.01
)
// Helper Functions:
// @function Returns the ratio to max/min of a sample period
// @param src float, data source.
// @param length int, period of the sample.
// @returns [float, float] tuple.
volatility (float src, int length) =>
[(src / ta.highest(src, length)[1] - 1.0) * 100.0,
(src / ta.lowest (src, length)[1] - 1.0) * 100.0]
// @function Volume Flow Indicator.
// @param Period int, period of the data sample.
// @param VCoef float, Volume Volatility Coefficient.
// @param Coef float, Cutoff Coefficient.
// @returns float.
// ref: https://mkatsanos.com/volume-flow-vfi-indicator/
vfi (int Period = 130, float VCoef = 2.5, float Coef = 0.2) =>
lastHLC3 = nz(hlc3[1], hlc3)
MF = hlc3 - lastHLC3
Vinter = ta.stdev(math.log(hlc3) - math.log(lastHLC3), 30)
Vave = ta.sma(volume, Period)[1]
Cutoff = Coef * close * Vinter
VC = math.min(volume, Vave * VCoef)
VCP = MF > Cutoff ? VC :
MF < -Cutoff ? -VC : 0.0
VFI1 = nz(math.sum(VCP, Period) / Vave)
VFI = ta.ema(VFI1, 3)
// inputs:
// optional strategy obserservation window parameters:
string ig_ow = 'Observation Window:'
bool i_Sdate = input( title = 'Start date:',
defval = timestamp('2021-01-01'),
inline = 'Sdate',
group = ig_ow
) < time //
bool i_useSdate = input.bool( title = '',
defval = false,
group = ig_ow,
inline = 'Sdate',
tooltip = 'Optional start date to clamp strategy observation window.'
) //
bool i_Edate = input( title = 'End date:',
defval = timestamp('2022-01-01'),
inline = 'Edate',
group = ig_ow
) > time //
bool i_useEdate = input.bool( title = '',
defval = false,
group = ig_ow,
inline = 'Edate',
tooltip = 'Optional end date to clamp strategy observation window.'
) //
//
string ig_ro = 'Lookback Options:'
int i_lback = input.int( title = 'Lookback Shift:',
defval = 0, minval = 0,
group = ig_ro,
tooltip = 'Optional, inspect previous signal values.'
) //
//
string ig_so = 'Signal Options:'
bool i_onlyL = input.bool( title = 'Long Only:',
defval = true,
group = ig_so,
tooltip = 'If switched off, short entries are initiated by sell signals.'
) //
int i_sMonth = input.int( title = 'Sell Month:',
defval = 8, minval = 1, maxval = 12, step = 1,
group = ig_so,
tooltip = 'The worst performing month, originally clamped between months 5 and 8.'
) //
int i_maxVI = input.int( title = 'Max VIX up:',
defval = 60, minval = 50, maxval = 60, step = 5,
group = ig_so,
tooltip = 'Volatility maximum threshold.'
) //
int i_critVFI = input.int( title = 'Critical VFI Sell:',
defval = -20, minval = -20, maxval = -15, step = 5,
group = ig_so,
tooltip = 'Critical money float (VFI) threshold for sell signal.'
) //
float i_K = input.float( title = 'ATR/VIX Ratio:',
defval = 1.5, minval = 1.3, maxval = 1.7, step = 0.2,
group = ig_so,
tooltip = 'ATR to VIX ratio for sell signal.'
) //
//
string i_VIticker = input( title = 'Volatility Index:',
defval = 'VIX',
group = ig_so,
tooltip = 'Volatility Index Ticker.'
) //
string i_VItf = input.timeframe( title = '',
defval = 'D',
group = ig_so,
tooltip = 'Volatility Index Timeframe.'
) //
int i_VIiperiod = input.int( title = 'Implied Volatility period:',
defval = 25,
group = ig_so
) //
int i_VIhperiod = input.int( title = 'Historical Volatility period:',
defval = 15,
group = ig_so
) //
//
int i_VFIperiod = input.int( title = 'VFI period:',
defval = 130,
group = ig_so, inline = 'VFI1'
) //
int i_VFIMperiod = input.int( title = 'MA:',
defval = 10,
group = ig_so, inline = 'VFI1',
tooltip = 'VFI and Moving Average sampling period.'
) //
float i_VFIcoef = input.float( title = 'VFI Coef Cuttoff:',
defval = 0.2,
group = ig_so, inline = 'VFI2'
) //
float i_VFIvcoef = input.float( title = 'Volat.:',
defval = 2.5,
group = ig_so, inline = 'VFI2',
tooltip = 'VFI Cutoff and Volatility coefficient.'
) //
int i_ATRperiod = input.int( title = 'ATR length:',
defval = 15,
group = ig_so, inline = 'ATR',
tooltip = 'ATR length.'
) //
//
string ig_to = 'Table Options:'
bool i_showT = input.bool( title = 'Show Table:',
defval = false,
group = ig_to,
tooltip = 'Optional toggle.'
) //
string i_Tpos = input.string(title = 'Position:',
defval = position.middle_right,
options = [ position.top_left, position.top_center, position.top_right,
position.middle_left, position.middle_center, position.middle_right,
position.bottom_left, position.bottom_center, position.bottom_right ],
group = ig_to) //
int i_Ttransp = input.int( title = 'Transparency:',
defval = 0, minval = 1, maxval = 99,
group = ig_to
) //
//
color i_Tcframe = input.color( title = 'Table Colors:',
defval = #000000,
group = ig_to, inline = 'table color'
) //
color i_Tcrowe = input.color( title = '',
defval = #d6dae3,
group = ig_to, inline = 'table color'
) //
color i_Tcrowo = input.color( title = '',
defval = #cccccc,
group = ig_to, inline = 'table color',
tooltip = 'Table background colors, in order: frame, even row, odd row.'
) //
string i_Ttsize = input.string(title = 'Table Text:',
defval = size.small,
options = [size.auto, size.huge, size.large, size.normal, size.small, size.tiny],
group = ig_to, inline = 'table text'
) //
color i_Tcdeft = input.color( title = 'Text Colors:',
defval = #000000,
group = ig_to, inline = 'table text'
) //
color i_Tcsigt = input.color( title = '',
defval = color.red,
group = ig_to, inline = 'table text'
) //
color i_Tctitt = input.color( title = '',
defval = color.navy,
group = ig_to, inline = 'table text',
tooltip = 'Table text size and colors, in order: default, short signal, title.'
) //
// Comparison Index
float VIX = request.security(i_VIticker, i_VItf, close)
[VIdn, VIup] = volatility(VIX, i_VIiperiod) // Implied
[ATRdn, ATRup] = volatility(ta.atr(i_VIhperiod), i_VIiperiod) // Historical
float VFI = vfi(i_VFIperiod, i_VFIvcoef, i_VFIcoef)
float VFI10 = ta.sma(VFI, i_VFIMperiod)
//
bool VFIatCrit = VFI > i_critVFI
bool lowVolat = (VIup < i_maxVI) or (ATRup < (i_K * i_maxVI))
bool VolatC = VFIatCrit ? lowVolat : false
bool Long = ((month >= 10) or (month < i_sMonth)) and VolatC[1]
bool Sseasonal = month == i_sMonth // SEASONAL EXIT/SHORT
bool Svol = VIup > (2.0 * i_maxVI) // VOLATILITY EXIT/SHORT
bool Scrit = ta.cross(i_critVFI, VFI) and (VFI10 < VFI10[1]) // VFI EXIT/SHORT
bool Short = Sseasonal or Svol[1] or Scrit[1]
bool withinObsWindow = true
//
if withinObsWindow and strategy.equity > 0
_L = strategy.long
_S = strategy.short
strategy.entry('L' , direction = _L, when = Long )
if i_onlyL
strategy.close('L', comment = 'EXIT SEASONAL' , when = Sseasonal )
strategy.close('L', comment = 'EXIT VOLATILITY', when = Svol[1] )
strategy.close('L', comment = 'EXIT MF' , when = Scrit[1] )
else
strategy.entry('S Seasonal' , direction = _S, when = Sseasonal )
strategy.entry('S Volatility', direction = _S, when = Svol[1] )
strategy.entry('S MF Crit.' , direction = _S, when = Scrit[1] )
else
strategy.close_all()
string SIGNAL = switch
(Long) => 'Long Seasonal'
(Sseasonal and i_onlyL) => 'Exit Seasonal'
(Svol[1] and i_onlyL) => 'Exit Volatility'
(Scrit[1] and i_onlyL) => 'Exit Money Flow'
(Sseasonal and not i_onlyL) => 'Short Seasonal'
(Svol[1] and not i_onlyL) => 'Short Volatility'
(Scrit[1] and not i_onlyL) => 'Short Money Flow Bearish'
=> 'none'
string date = str.format(
'{0,number,0000}-{1,number,00}-{2,number,00}',
year, month, dayofmonth
)
var table dTable = table.new(position = i_Tpos,
columns = 2,
rows = 17,
frame_color = color.new(#000000, i_Ttransp),
frame_width = 4
) //
// @function Helper to populate the table rows.
tRow(tableId, idx, left, right, tcol=0) =>
color _bg = color.new(idx % 2 ? i_Tcrowo : i_Tcrowe, i_Ttransp)
color _tx = switch (tcol)
(1) => color.new(i_Tcsigt, i_Ttransp)
(2) => color.new(i_Tctitt, i_Ttransp)
=> color.new(i_Tcdeft, i_Ttransp)
// table.cell( table_id=tableId,
// column=0, row=idx,
// text=left, text_color=_tx, text_halign=text.align_right, text_size=i_Ttsize,
// bgcolor=_bg) //
// table.cell( table_id=tableId,
// column=1, row=idx,
// text=str.tostring(right), text_color=_tx, text_halign=text.align_left, text_size=i_Ttsize,
// bgcolor=_bg) //
if i_showT
float _atr10 = ta.atr(10)[i_lback]
string _nf = '0.00'
string _aru = '? ', string _ard = '? '
// id | idx | left label | right label | conditional color |
tRow(dTable, 00, 'S&P500 Hybrid Seasonal ' , '' , 2 )
tRow(dTable, 01, 'Created By: Markos Katsanos' , '' , 2 )
tRow(dTable, 02, 'Date:' , date[i_lback] )
tRow(dTable, 03, 'Signal:' , SIGNAL[i_lback] )
tRow(dTable, 04, 'Price:' , open[i_lback] )
tRow(dTable, 05, 'VIX:' , str.tostring( VIX[i_lback], _nf) )
tRow(dTable, 06, 'VFI:' , str.tostring( VFI[i_lback], _nf) , VFIatCrit ? 1 : 0 )
tRow(dTable, 07, 'ATR:' , str.tostring( _atr10, _nf) )
tRow(dTable, 08, 'VIup%:' , str.tostring( VIup[i_lback], _nf) , VIup > i_maxVI ? 1 : 0 )
tRow(dTable, 09, 'ATRup%:' , str.tostring(ATRup[i_lback], _nf) , ATRup > i_K * i_maxVI ? 1 : 0 )
tRow(dTable, 10, 'VIdn%:' , str.tostring( VIdn[i_lback], _nf) )
tRow(dTable, 11, 'ATRdn%:' , str.tostring(ATRdn[i_lback], _nf) )
tRow(dTable, 12, _aru + 'Long Seasonal:' , Long[i_lback] )
tmp = 12
if not i_onlyL
tmp := 13
tRow(dTable, 13, _ard + 'Short:' , Short[i_lback] , Short[i_lback] ? 1 : 0 )
tRow(dTable, tmp+1, _ard + 'Seasonal:' , Sseasonal[i_lback] , Sseasonal[i_lback] ? 1 : 0 )
tRow(dTable, tmp+2, _ard + 'Volatility:' , Svol[1+i_lback] , Svol[1 + i_lback] ? 1 : 0 )
tRow(dTable, tmp+3, _ard + 'Money Flow:' , Scrit[i_lback] , Scrit[i_lback] ? 1 : 0 )
Published by ChaoZhang on the FMZ strategy square · last modified 2024-01-22. Collected in the Quant Trading Vault. Unvalidated research: re-test with your own data and costs before trusting it.
Build this with AgenKit
A spec is not a trading system. Paste this into your coding agent and AgenKit runs a specialist team — architecture, test-first build, risk gates, review — with paper trading on by default.
/agenkit build a production-grade trading bot with a backtest engine, risk controls and paper trading by default from strategies/SP500-Hybrid-Seasonal-Trading-Strategy.md- npx agenkit install engineering-kit in your project
- Paste the command into your agent and approve each phase
- Get tests, backtest, risk controls and a paper-trading switch
Understand it first
Before you run this, read the lesson behind the idea and how to backtest it honestly. Specs like this one are the survivors of an unknown number of trials; deflate their results accordingly.
Related specs
- Dual Indicator Stochastic RSI and EMA Trading StrategyPine Script · Machine learning · 1d
- ADX EFI 50 Moving Average Channel Pullback StrategyPine Script · Machine learning · 1h
- Multi indicator Combination StrategyPine Script · Machine learning · 1d
- AI Driven Dynamic Volatility Adaptive Trend Breakout Trading StrategyPine Script · Machine learning · 2d
- Dual Moving Average Matching Strategy Based on Bollinger BandsPine Script · Machine learning · 1h
- High Frequency Hybrid Technical Analysis Quantitative StrategyPine Script · Machine learning · 15m