ExponentialSmoothingRegressor predicts horizontal line for yahoo data

#### Code Sample, a copy-pastable example to reproduce your bug.

```python
import pandas as pd
from evalml.pipelines.components import ExponentialSmoothingRegressor
from evalml.preprocessing import split_data

X = pd.read_csv("/Users/freddy.boulton/Downloads/yahoo_stock.csv")
y = X.pop("Close")
X = X[["Date"]]

problem_configuration = {"gap": 0, "max_delay": 50, "forecast_horizon": 20,
"time_index": "Date"}

X_train, X_valid, y_train, y_valid = split_data(X, y, "time series regression", problem_configuration)

esr = ExponentialSmoothingRegressor()
esr.fit(X_train, y_train)
esr.predict(X_valid).plot()
```

[yahoo_stock.csv.zip](https://github.com/alteryx/evalml/files/7990830/yahoo_stock.csv.zip)