The Research Behind the Model

How We Build Monthly Allocation Guidance

Our model draws on decades of academic research in momentum investing, trend following, and behavioral finance to construct a rules-based allocation system.

01

The Foundation of Our Approach

For 401k users, one of the biggest questions is "What should I invest in"? In the past the default option was cash, earning almost no interest. But more recently, target date funds have become the default option. But they have a flaw - as people live longer, they need their money to work longer. Target date funds automatically shift into bonds. But bonds are no longer the buffer they once were. When long term interest rates trend up, bonds lose value. You also lose out to potential gains in the stock market during strong uptrends.

02

Momentum is your friend

Decades of academic research — from Jegadeesh & Titman to Asness, Moskowitz & Pedersen — document that assets with strong relative performance over the past 3–12 months tend to continue outperforming over the next 1–6 months. Our model measures cross-asset momentum monthly and tilts allocations accordingly.

03

Trend Following for Drawdown Control

401k investors are in it for the long term, and that's why being on the right side of the trend for the majority of the time is important. We use trend filters to cut equity exposure during sustained downtrends — systematically, without emotion, every month.

04

We reduce risk when it gets stormy

Market volatility is like the bodies pulse. When things are running smooth, the pulse is normal. When volatility is elevated that can signal something is wrong. We adjust risk by reducing market exposure when things get volatile.

05

Monthly Rebalance Cadence*

Rebalancing too frequently generates excess trading; too infrequently lets drift erode the model's signal. Monthly cadence balances responsiveness with stability — and maps cleanly to the rebalance windows most 401(k) plans allow. *The exception is during Crashguard warning periods, where we'll check twice a month.

06

Our strategy vs buy and hold.

If you were close to retirement at the beginning of 2000, you would had to endure a brutal 50% market drawdown, and it took 13 years for it to recover back to its former high. Our method would have limited the downside to less than 14% with essentially the same long term return. In 2000, the market was severely overvalued, and we find ourselves in the same position now. We can't predict or avoid the next major market downturn, but we can minimize risk if and when it does come.

Track Record

Annual Performance

Annual returns from 2000 through 2026, covering two major market crises. The model navigated the dot-com bust and the 2008 crash with limited drawdowns while capturing the bulk of upside in strong years.

+32.3%

−10.1%

2000

+5.20%

2001

+0.08%

2002

-5.41%

2003

+21.26%

2004

+14.76%

2005

+4.18%

2006

+13.17%

2007

+5.15%

2008

-0.33%

2009

+8.56%

2010

-2.07%

2011

-2.56%

2012

+7.88%

2013

+32.31%

2014

+13.46%

2015

-4.26%

2016

+7.34%

2017

+21.71%

2018

-10.08%

2019

+10.95%

2020

+14.58%

2021

+23.71%

2022

-6.26%

2023

+10.59%

2024

+24.17%

2025

+11.46%

2026

+4.95%

2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
2021
2022
2023
2024
2025
2026

20

Positive years

8.3%

Average annual return

-10.1%

Worst year

$10,000 Invested in 2000

Cumulative portfolio growth — model vs. S&P 500 buy & hold. Hover to inspect any year.

$5927 $24.5k $43.1k $61.7k $80.3k 2000 2004 2008 2012 2016 2020 2024 2026
401kToolworks Model
S&P 500 Buy & Hold

$76.5k

Model — ending value (2026)

$62.1k

S&P 500 — ending value (2026)

Drawdown From Peak — Model vs. S&P 500

Running drawdown from the prior equity peak each year. Shallower troughs mean less capital destruction during downturns. Hover to inspect any year.

0% -10% -20% -30% -41% 2000 2004 2008 2012 2016 2020 2024 2026
401kToolworks Model
S&P 500 Buy & Hold

-10.1%

Model — worst drawdown

-37.6%

S&P 500 — worst drawdown

Past performance is not indicative of future results. Returns represent backtested model output and should not be construed as a guarantee of future performance.