---
title: Smooth Sign Accuracy (SSA) Framework
url: https://www.emergentmind.com/topics/smooth-sign-accuracy-ssa
type: topic
---

# Smooth Sign Accuracy (SSA) Framework

Smooth Sign Accuracy (SSA) is a forecasting and signal-extraction framework designed for settings in which conventional mean-squared-error (MSE) optimization is too narrow. Its central premise is that a useful predictor should not only track a target in level, but should also be directionally reliable and should avoid excessive, spurious sign changes. SSA formalizes this requirement as an **accuracy-smoothness (AS) dilemma**: improving tracking accuracy typically moves a predictor toward the classical MSE solution, which can be noisy, while enforcing smoothness reduces false turning points at some cost in target tracking. The framework therefore maximizes predictive alignment with a target subject to an explicit smoothness constraint, where smoothness is measured by the expected time between zero crossings, or equivalently by the frequency of sign changes [2601.06547]. Its multivariate extension, M-SSA, incorporates cross-sectional information across several time series and is used for forecasting, nowcasting, and smoothing on an accuracy-smoothness frontier [2602.13722].

## 1. Formal criterion and basic setup

SSA is formulated for a linear prediction problem. The target is written as
$$
z_t := \sum_{k=-\infty}^{\infty} \gamma_k x_{t-k},
$$
and the causal linear predictor of \(z_{t+\delta}\) is
$$
y_t=\sum_{k=0}^{L-1} b_k x_{t-k},
$$
where \(\delta>0\) denotes forecasting, \(\delta=0\) nowcasting, and \(\delta<0\) backcasting [2601.06547].

The theory is first developed under \(x_t=\epsilon_t\), i.i.d. white noise, with \(\epsilon_t\) standardized. In that case,
$$
y_t=\mathbf b' \boldsymbol{\epsilon}_t,\qquad \boldsymbol{\epsilon}_t=(\epsilon_t,\ldots,\epsilon_{t-(L-1)})',
$$
and the classical MSE predictor is
$$
y_{t,MSE}=\boldsymbol{\gamma}_\delta' \boldsymbol{\epsilon}_t,
$$
with \(\boldsymbol{\gamma}_\delta=(\gamma_\delta,\ldots,\gamma_{\delta+L-1})'\) [2601.06547].

The core SSA criterion fixes a smoothness level and then maximizes predictive alignment with the target. In equivalent form, it is
$$
\max_{\mathbf b}\rho(y,z,\delta)
\quad\text{subject to}\quad
\rho(y)=\rho_1,\qquad
\mathbf b'\mathbf b=l,
$$
where
$$
\rho(y)=\frac{\mathbf b' \mathbf M \mathbf b}{\mathbf b'\mathbf b}
$$
is the lag-1 autocorrelation of the predictor, and \(\mathbf M\) is the tridiagonal matrix with \(0.5\) on the first off-diagonals, so that
$$
\mathbf b' \mathbf M \mathbf b=\sum_{k=1}^{L-1} b_{k-1}b_k.
$$
The framework denotes this solution by \(\mathrm{SSA}(\rho_1,\delta)\) [2601.06547].

This formulation generalizes MSE-based prediction rather than replacing it. Under white noise, the MSE predictor is recovered by choosing
$$
l=\boldsymbol{\gamma}_\delta'\boldsymbol{\gamma}_\delta,\qquad
\rho_1=\frac{\boldsymbol{\gamma}_\delta' \mathbf M \boldsymbol{\gamma}_\delta}{\boldsymbol{\gamma}_\delta'\boldsymbol{\gamma}_\delta} =: \rho_{MSE}.
$$
The paper also gives an MSE variant of SSA that minimizes squared error subject to the holding-time restriction, making the relationship to classical MSE explicit [2601.06547].

## 2. Sign accuracy, holding time, and the accuracy-smoothness dilemma

SSA is built on two Gaussian identities that connect correlation, directional performance, and smoothness. For a zero-mean stationary Gaussian predictor \(y_t\), sign accuracy is defined as
$$
SA(y_t)=P(y_t z_{t+\delta}>0),
$$
and satisfies
$$
SA(y_t)=0.5+\frac{\arcsin(\rho(y,z,\delta))}{\pi}.
$$
Because \(\arcsin(\cdot)\) is strictly increasing on \([-1,1]\), maximizing target correlation is equivalent to maximizing sign accuracy [2601.06547].

Smoothness is defined through the expected duration between successive zero crossings, the **holding time**
$$
ht(y|\mathbf b)=\frac{\pi}{\arccos(\rho(y))}
$$
for stationary Gaussian \(y_t\). A higher \(\rho(y)\) therefore means a larger holding time, fewer sign changes, and a smoother predictor [2601.06547]. This is a distinct notion of smoothness: SSA does not penalize curvature in levels, but regulates the expected zero-crossing rate.

The AS dilemma is the statement that these objectives conflict. On the relevant solution branches, increasing smoothness lowers target correlation and hence lowers sign accuracy; decreasing smoothness does the opposite [2601.06547]. A dual interpretation follows: SSA can be read not only as the most accurate predictor at a chosen smoothness level, but also as the smoothest predictor among all linear predictors with a specified target correlation [2601.06547].

This gives \(\rho_1\), or equivalently
$$
ht_1=\frac{\pi}{\arccos(\rho_1)},
$$
a direct behavioral interpretation. Raising \(\rho_1\) suppresses false turning points and enforces a more monotone trajectory; lowering it makes the predictor more reactive but noisier [2601.06547].

## 3. Parametric solution and filter interpretation

A central analytical result is that the problem has a one-parameter characterization. The matrix \(\mathbf M\) has eigenvalues
$$
\lambda_j=\cos(\omega_j),\qquad \omega_j=\frac{j\pi}{L+1},\quad j=1,\ldots,L,
$$
so the feasible smoothness range is
$$
-\cos\!\left(\frac{\pi}{L+1}\right)\le \rho(y)\le \cos\!\left(\frac{\pi}{L+1}\right).
$$
Thus the admissible first-order autocorrelation is bounded by the extremal eigenvalues of \(\mathbf M\) [2601.06547].

If the MSE filter is written spectrally as
$$
\boldsymbol{\gamma}_{\delta}=\sum_{i=n}^{m}w_i\mathbf{v}_i,
$$
then, under the stated regularity conditions, the SSA solution has the form
$$
\mathbf{b}(\nu)=D(\nu,l)\mathbf{N}^{-1}\boldsymbol{\gamma}_{\delta}
= D(\nu,l)\sum_{i=1}^L \frac{w_i}{2\lambda_{i}-\nu}\mathbf{v}_{i},
$$
with
$$
\mathbf N:=2\mathbf M-\nu \mathbf I.
$$
The unknown scalar \(\nu\) is chosen so that the induced lag-1 autocorrelation equals the prescribed \(\rho_1\) [2601.06547].

In the time domain, \(\mathbf b(\nu)\) satisfies the reversible second-order difference equation
$$
b_{k+1}(\nu)-\nu b_k(\nu)+b_{k-1}(\nu)=D\gamma_{k+\delta},
$$
with boundary conditions
$$
b_{-1}(\nu)=b_L(\nu)=0.
$$
This AR(2)-like representation gives the paper’s main intuition: SSA smooths the target by convolving it with an AR(2)-type filter whose single free parameter \(\nu\) controls smoothness [2601.06547].

The frequency-domain transfer function is
$$
\Gamma_{AR(2)}(\nu,\omega)=\frac{1}{2\cos(\omega)-\nu}.
$$
Its interpretation is direct. If \(\nu\ge 2\), the induced filter is low-pass and suppresses high-frequency noise; if \(\nu\le -2\), it is high-pass and increases sign changes; if \(-2<\nu<2\), it is band-pass [2601.06547]. In that sense, \(\rho_1>\rho_{MSE}\) corresponds to deliberate smoothing relative to MSE, while \(\rho_1<\rho_{MSE}\) corresponds to deliberate unsmoothing.

## 4. Dependent and integrated processes

SSA is extended beyond white noise through the Wold decomposition
$$
x_t=\sum_{i=0}^{\infty}\xi_i\epsilon_{t-i},
$$
with invertible MA representation. The criterion is then solved in the innovation domain and transformed back by deconvolution [2601.06547]. A practical consequence emphasized in the paper is that, unlike a fixed benchmark filter whose effective smoothness depends on the data-generating process, SSA maintains the chosen holding time across processes.

The most distinctive extension concerns integrated processes. For \(\tilde x_t\) that is \(I(1)\), correlation and zero-crossing rates in levels are not well-defined in the same way, so the framework controls sign changes of the **first differences** of the predictor instead. The relevant object is the error relative to the level MSE predictor,
$$
e_{SSA,t}:=y_{t,MSE}-y_t,
$$
together with a cointegration constraint ensuring stationarity of this error [2601.06547].

In the \(I(1)\) case, the emphasized MSE-based version imposes the holding-time restriction on
$$
\mathbf b_\epsilon' \boldsymbol\epsilon_t \approx y_t-y_{t-1},
$$
so \(\rho_1\) controls sign changes in the predictor’s first differences. The resulting level predictor is described as **maximal monotone**: among predictors with the same MSE-type tracking performance, it minimizes sign changes of \(y_t-y_{t-1}\) [2601.06547]. For \(I(2)\) processes, the same logic is applied to second differences, yielding predictors with the **fewest inflection points / lowest curvature** in levels [2601.06547].

Implementation is correspondingly structured. The paper summarizes it as: compute the benchmark MSE predictor, choose the desired smoothness level via \(\rho_1\) or \(ht_1\), solve for \(\nu\) on the monotone branch in the stationary case, transform through MA inversion for dependent processes, and impose cointegration for integrated processes [2601.06547].

## 5. Multivariate extension: M-SSA

The multivariate extension, M-SSA, lets the predictor for one target series use the full multivariate system. With
$$
\mathbf{x}_t = (x_{1t}, \ldots, x_{nt})',
$$
a multivariate Wold decomposition
$$
\mathbf{x}_t = \sum_{k=0}^\infty \boldsymbol{\Xi}_k \boldsymbol{\epsilon}_{t-k},
$$
target
$$
\mathbf{z}_t=\sum_{|k|<\infty}\boldsymbol{\Gamma}_k\mathbf{x}_{t-k},
$$
and predictor
$$
\mathbf{y}_t = \sum_{k=0}^{L-1} \mathbf{B}_k \mathbf{x}_{t-k},
$$
the white-noise version of the criterion is expressed with
$$
\tilde{\mathbf I}:=\boldsymbol{\Sigma}\otimes\mathbf I_{L\times L},\qquad
\tilde{\mathbf M}:=\boldsymbol{\Sigma}\otimes\mathbf M
$$
and, for each target component \(i\),
$$
\max_{\mathbf b_i}\boldsymbol{\gamma}_{i\cdot\delta}'\tilde{\mathbf I}\mathbf b_i
\quad\text{subject to}\quad
\mathbf b_i'\tilde{\mathbf M}\mathbf b_i=\rho_i,\qquad
\mathbf b_i'\tilde{\mathbf I}\mathbf b_i=1.
$$
This is the direct multivariate analog of the univariate SSA problem [2602.13722].

Cross-sectional information enters through the covariance matrix \(\boldsymbol{\Sigma}\), the stacked target coefficients, and the multivariate Wold representation. For target \(i\), the predictor is built from lagged values of all \(n\) series, not just its own history [2602.13722]. The practical implication is that smoother or leading series can improve the target’s nowcast or forecast while preserving a prescribed holding time.

The M-SSA solution again has a one-parameter form,
$$
\mathbf{b}_{i}=D_i\mathbf{N}_i^{-1}\tilde{\mathbf I}\boldsymbol{\gamma}_{i\cdot\delta},
\qquad
\mathbf{N}_i:=2\tilde{\mathbf M}-\nu_i\tilde{\mathbf I},
$$
with \(\nu_i\) chosen to satisfy the desired autocorrelation constraint [2602.13722]. In time-domain form, the coefficients satisfy a non-stationary and time reversible difference equation with zero boundary conditions.

The framework is used in three application domains. For forecasting, the target is a future value. For nowcasting, \(\delta=0\) and the target may be an acausal signal such as a two-sided trend filter. For smoothing, the target is causal, and M-SSA acts as a smoother whose smoothness notion differs from Whittaker-Henderson or Hodrick-Prescott smoothing because it controls zero-crossing frequency rather than curvature [2602.13722].

## 6. Empirical behavior and comparative results

The empirical case studies are intended to show how the AS frontier behaves in practice. In a quarterly \(\mathrm{HP}(1600)\) customization with \(L=101\), the MSE nowcast has \(\rho_{MSE}=0.926\). Two alternative filters are reported: \(\mathrm{SSA}(0.97,0)\), which smooths relative to MSE, and \(\mathrm{SSA}(0.8,0)\), which unsmooths. Their holding times are \(8.138\) for MSE, \(12.793\) for \(\mathrm{SSA}(0.97,0)\), and \(4.882\) for \(\mathrm{SSA}(0.8,0)\), while the target HP filter has holding time \(34.316\) [2601.06547].

In the monthly U.S. industrial production example, modeled as approximately ARIMA(1,1,0) after log differencing, the target is a two-sided monthly HP trend with \(\lambda=14400\). The comparison is among the level MSE predictor, the classic one-sided HP concurrent filter, and the proposed \(I(1)\)-SSA predictor. Reported in-sample MSEs relative to the two-sided HP target are \(0.00024\) for MSE, \(0.00037\) for \(I(1)\)-SSA, and \(0.00050\) for HP-C; empirical holding times of differenced predictors are \(2.29630\), \(25.83333\), and \(18.23529\), respectively [2601.06547]. The stated interpretation is that \(I(1)\)-SSA materially improves smoothness over MSE and improves accuracy over HP-C.

For multivariate nowcasting, the bivariate INDPRO + CLI example is the clearest illustration of how leading-indicator information changes the trade-off [2602.13722].

| Method | Target corr. | HT |
|---|---:|---:|
| HP-C | 0.650 | 11.132 |
| M-SSA | 0.736 | 17.263 |
| MSE | 0.744 | 11.011 |

These results are paired with the qualitative claim that M-SSA achieves the desired 50% holding-time increase over MSE, loses only a little target correlation relative to MSE, beats HP-C on both target correlation and holding time, and eliminates the lag observed for univariate SSA in this application [2602.13722].

A further contrast with curvature-based smoothers appears in the white-noise smoothing comparison with HP. With \(L=201\) and \(\rho_1=0.9986\), SSA1 matches HP’s holding time and attains higher target correlation, while SSA2 matches HP’s target correlation and attains a larger holding time; HP nevertheless remains superior in RMS second-order differences [2602.13722]. This supports the paper’s claim that SSA smoothness is specifically about sign-change frequency rather than curvature.

## 7. Terminological ambiguity of “SSA”

The abbreviation **SSA** is overloaded in the literature, and disambiguation is often necessary. In two time-series papers, it refers to **Singular Spectrum Analysis**, including work on structured Hankel implementations and on forecasting parameter selection [0911.4498; 2403.16507]. In sign-language processing, **SSA** denotes **sign–subtitle alignment**, a temporal localization task over signing video rather than a forecasting criterion [2512.08040]. In optimization, smooth sign transformations have been studied for sign-based optimizers, but that work does **not** define a concept called Smooth Sign Accuracy [2605.31371].

Within the forecasting literature, by contrast, **Smooth Sign Accuracy** denotes the criterion that maximizes target correlation—and, under Gaussianity, sign accuracy—subject to explicit control of the predictor’s sign-change rate through holding time [2601.06547]. Its multivariate extension, M-SSA, preserves that definition while allowing each target to borrow information from multiple series [2602.13722].

Source: https://www.emergentmind.com/topics/smooth-sign-accuracy-ssa